The demand by Asian retail investors to trade US stocks during Asia’s daytime is pushing the US itself to extend its own trading hours to be open nearly around the clock. The longer term implications, should trading equities in tokenized form become prevalent, and which implies a 24/7 market, is that global equities markets are being pushed closer to a crypto-like environment, even if the drivers of this trend are unrelated.

The US equities market is the world’s largest, with a total market capitalization of $66 trillion. Daily trading volumes continue to reach record highs, hitting 17.6 billion shares and notional value of more than $1 trillion.

Investors in Asia and the Middle East have long desired to trade this market in their own daytimes, a wish that has recently become a reality. The boom in memestocks and raging tech names such as Nvidia and Tesla have added to the furore.

Early ATS

The first attempts emerged in Japan as long ago as 1998, when the first alternative trading systems launched.

ATSs are broker-dealers that share the trading mechanics, technology, and participants as stock exchanges. The difference is regulation. Exchanges are self-regulated organizations with listing standards and surveillance obligations. ATSs are broker-dealers supervised by securities regulators and with fewer disclosure requirements.

Japan labels its version of these ‘proprietary trading systems’. There were 10 PTSs active during the 2000s, but they failed to crack the market. Today there are now just two (Japannext, owned by SBI, and Osaka Digital Exchange; a third, Cboe Japan, closed in 2025 with Chicago-based Cboe taking stakes in the two survivors). It has taken that industry 28 years to achieve 12 percent market share in Japan.

Its impact in the US has been minimal: these PTSs are only allowed to trade in Japanese stocks. But they established the idea of ‘night trading’, allowing Japanese punters to observe US and European market activity before placing trades in Japan, with PTSs operating sessions in the evening.

Blue Ocean in Korea

Although these PTSs don’t touch US equities, they highlighted interest among Asian retail investors to trade after work and after dinner, and the fact that they actively tracked US markets.

In June 2021, a new US-based ATS called Blue Ocean connected the dots and pitched itself directly to Korean retail investors, offering to place trades in US stocks during the Korean day. Blue Ocean’s innovation was to create a new trading schedule in the US, operating from Sunday 8pm to Thursday 4am, US Eastern Standard Time.

That corresponds to daytime in Korea, and Korean retail makes up 40 percent of Blue Ocean’s volumes; other Asian markets comprise another 30 percent, while American investors who want to trade during their nighttime (especially meme stocks) make up the remaining 30 percent. Blue Ocean dominated US overnight trading volume, which has come to make up about 1 percent of total US equity volume.

The experiment almost collapsed. Blue Ocean relies on local Korean brokers, notably including Samsung Securities, to reach their investors. Rival brokers and Korean regulators were alarmed by the huge surge in uptake; Blue Ocean also suffered a system outage that led to questions about investor protections. Korean brokers suspended Blue Ocean’s service in 2024, and it only resumed in early 2025; the ATS has since opened an office in Seoul.

NexTrade

The wild success of Blue Ocean convinced South Korean regulators that they needed to upgrade their own market, lest they see more volume sucked into US firms. In 2025, shortly before Blue Ocean was allowed to resume service, the Korean Securities and Futures Commission broke Korea Exchange’s seven-decades long monopoly on stock exchange business by granting a license to the first homegrown ATS, NexTrade.

Like its Japanese peers, NexTrade only allows trading of domestic stocks, and it differentiated itself by offering extended trading hours into the Korean evening, along with lower fees. Unlike in Japan, the authorities in Seoul did not burden NexTrade with many rules designed to protect incumbent exchanges.

NexTrade took off like a rocket. It took it less than one year to secure 30 percent of the local market share, compared to the 12 percent that Japan’s PTSs have eked out in their market after more than 25 years.

At the same time, two more American ATSs have entered the Asian markets: Moon ATS, operated by OTC Markets Group, and Bruce ATS, operated by Bruce Markets, a company operated by PEAK6 Investments. Both ATSs were licensed in the US in late 2024, and explicitly target retail investors in Asia. Both were able to sign distribution deals with Korean domestic brokers, who were eager to break Blue Ocean’s monopoly as it resumed activity.

Going 23/5

These three US-based ATSs are now bringing liquidity to the US overnight market. Their flow is entirely retail, but that attracts market makers. These ATSs expect this will lead to interest from ETF brokers and designers of structured products. Should they realize their hopes of collectively winning up to 10 percent of total US equity volumes over the coming years, this amount of liquidity would finally encourage US institutional investors to participate in the overnight market.

Something like that may be bound to happen anyway. The rise of overnight trading in the US and the obvious appetite from Asian retail investors has pushed the US to transform its market to trade “23/5”, five days a week around the clock, with only a one-hour gap (at 8pm, EST) to reset the matching engines.

This new extended trading schedule is meant to launch on December 6 this year, pending full regulatory approvals.

Most market participants in the US don’t care about 23/5. They didn’t ask for it and it represents a new reality that will require thinking about risk management and operational flows.

But it’s not for US investors; it’s for Asians. For now, this market is dominated by three young ATSs, but when the US goes 23/5, the bulge bracket will jump in, offering access to a greater number of company stocks, including American depository receipts (synthetic representations of foreign-listed stocks).

Some participants also believe that more liquidity and attention will spur activity among US investors. Just as Asians want to trade their local market on the back of news out of the West, US investors may want to watch what happens in Asia to decide how to trade at home; the surprise announcement of DeepSeek in 2024 is one example of a market-moving event.

Building on T+1

The US decision to embrace a 23/5 system is not just a reaction to Asian demand. It is made possible by America’s own move to T+1 (settling one day after a trade) in 2024. This has compressed time allowed for brokers, asset managers, and custodians match, report, and close trades. So far the move from T+2 has gone smoothly, suggesting the processes are in place to handle additional flows during late hours.

The overnight ATSs have also impacted market efficiency. Spreads are narrowing. Market makers are now deploying smart-order routing to the ATSs, to enable best execution. (In the US, brokers are required to place an order at the exchange offering the best price, which is meaningful in such a big market with multiple stock exchanges; ATSs as broker-dealers are exempt from this rule, but competitive forces will probably enforce best-execution practices.)

Smart-order routing is a technology consideration. Moving to 23/5 requires some more tech upgrades and problems to be solved. It’s a big project, but it’s one that is well understood.

Challenges

There are bigger questions, mostly in the realm of tax and compliance, particularly regarding clients based overseas. These legalistic details will determine basic questions such as whether bulge-bracket firms serve Asian clients from teams in the US working the night shift, or if they place relationship managers (who are licensed with US SEC Series 7 credentials) in Asia.

There’s also going to be a mind shift: when the US working week begins on Sunday night, and the actual trading takes place “tomorrow” (crossing the international date line), firms will face personnel and management challenges.

Operations will pose more concrete issues for the industry. Some aspects of post-trade processing should remain straightforward, such as batching overnight trades to the following day’s work to calculate NAVs. But others are not. Corporate actions pose new risks when they involve many time zones. For now, the ATSs say they suspend trading in any ticker that announces a material action. This doesn’t provide the same assurance as the regulatory requirements on exchanges.

Another aspect that will need to be harmonized is order limits. When a stock’s price crashes (or surges), regulatory circuit-breakers force exchanges to freeze trading. ATSs don’t have to.

ATS executives say competitive forces will lead to harmonized rules, but without regulation, the ATSs will always be able to wriggle out of such obligations; which is why the big exchanges in the US may demand ATSs fall more under similar rules.

Retail investors are either oblivious to these details, or powerless to change them, but institutional investors have clout. It’s possible that regulation needn’t be heavy-handed, if overnight trading grows big enough to attract institutions, starting with hedge funds. Institutions would insist on tighter rules to protect themselves. That may lead to the necessary changes; if not, regulators will have to step in.

Regulation may or not be needed, but automation is going to be a must, if the markets are to maintain true straight-through processing 23/5. Ditto as more jurisdictions, including Europe and maybe Hong Kong, also move to T+1.

From 23/5 to 24/7?

While 23/5 is a big structural change to US equity markets, it might be just another step towards matching crypto’s 24/7 environment. NYSE and Nasdaq have already expressed interest in tokenizing equity assets. In January, NYSE said it will develop a blockchain-based paltform for trading and on-chain settlement of tokenized securities, which will operate as an ATS. The exchanges are already preparing to move beyond 23/5, in a phased approach.

The challenge isn’t tech: the DeFi world already supports fractionalized trading and round-the-clock trading. The challenge is how to explain this new world to the listed companies, whose leadership and investor-relations teams aren’t involved in crypto-anything, and could end up surprised by sudden volatility in their stock on a Sunday.

The best argument for caution is that people aren’t robots; we need a break. Technology and sensible rules around risk may make 24/7 – or even just 23/5 – possible. The direction seems clear to industry participants. The US is moving to 23/5 largely in response to Asian retail demand for overnight trading. On top of this, Wall Street firms are keen to adopt tokenization for its efficiencies, speed, and capital treatment. A non-stop market promises enormous volumes; it also guarantees new risks and new problems. Gradually, as overnight trading diversifies to include more institutional investors, it will serve as a stepping stone to enable at least part of the market to dance all night long.

Ned is an ex-banker, programmer at a hedge fund, and former CTO of SingLife during its time as an independent insurtech. He also shares his insights about AI on LinkedIn. Given this background, he’s in a good position to help execs in the financial world get a better handle on what agentic AI means for them. Ned brings a view that straddles incumbents and startups, which is also useful for this conversation. Today he runs his own startup, Mission+, an agentic AI engineering company.

The central proposition is that firms need to understand when to use AI as a product - a thing they prompt to get some kind of output - and when it’s a tool to build a product, which can mean an incredibly productive programmer. None of this is straightforward but it’s all pretty exciting.

We framed this conversation deliberately around AI’s potential. We don’t get into issues around security and reliability, which seemed like a separate conversation. If you’re looking for reasons to shun AI or to focus on its (very real) shortcomings, this conversation isn’t for you. We spent our thirty minutes talking instead about the impact on financial firms, from a technologist steeped in the industry.

Timecodes:

0:00 - Ned Lowe, Mission+

2:48 - “AI as a product” versus “AI to build the product”

6:41 - Governance around ensuring genAI is reliable

8:58 - “Agile is dead” and the new digital transformation in finance

14:29 - Lessons from the trading floor and why AI restores the value of high touch

18:23 - Cost implications of AI, the cloud example, and how to think about cost vs benefits

23:25 - Is AI good or bad for financial incumbents grappling with legacy tech?

26:32 - Is AI going to enable fintechs to return to disruptive innovation?

29:11 - About Mission+

The launch of the eMPF Platform is a genuine public-policy and technology accomplishment. It is the most significant reform to Hong Kong’s Mandatory Provident Fund system since inception, replacing fragmented, trustee-specific administration with a centralized digital platform designed to reduce costs through standardization, automation, and streamlined processing. In April 2026, it completed uploading all member and industry accounts to its digital dashboard.

That accomplishment, however, should be qualified. Although on its own it is helpful to administrating the HK$1.55 trillion (US$198 billion) pension system, it is only transformative if viewed by policymakers, administrators, asset managers, and members as a beginning rather than an endpoint.

The rollout exposed real onboarding and service problems. Those might prove to be teething problems, but they have kept the eMPF Platform Company—established to administer the online service—on the defensive, rather than expressing a long-term vision. The official articulation of eMPF’s long-term purpose still remains narrower than the opportunity before Hong Kong: to turn a digitized pension utility into a more open, portable and innovation-friendly financial infrastructure that serves members better and deepens the city’s financial-services ecosystem.

What eMPF has already achieved

The introduction of eMPF is a foundational shift. Before this electronic platform was built (by PCCW, a local telco), MPF administration was fragmented across trustees and remained burdened by paper-heavy processes, high administrative costs, and weak member engagement. The platform created, for the first time, the possibility that members could see and manage their holdings across schemes through a single interface.​

According to the Mandatory Provident Fund Schemes Authority (MPFA), as of end-March 2026, more than 2 million users had registered on eMPF, including over 1.8 million scheme members and 200,000 employers, while the platform had processed more than 9 million administrative instructions, over 70% of them electronically.​

That is evidence that eMPF is not a cosmetic digital layer but a major operational infrastructure that is already changing how MPF is administered.​ Hong Kong’s working population is about 3.7 million, so if 2 million people have registered, that’s more than half.

Eric Lui, CEO of eMPF Platform Company (established to administer the online system), argues the service is already more ambitious than electronic dashboards used by foreign pension systems. Citing data from the Organization of Economic Cooperation and Development (OECD), Lui saysoverseas pension interfaces mainly display information, whereas eMPF combines visibility with transactional capability, including account opening, contributions, fund switching, changes to investment instructions, and account consolidation.​

Where the rollout fell short

Glass half full, or half empty? Nearly half the members haven’t registered. Many who did so may not become active users. The registration process was cumbersome (which is acceptable given security and privacy concerns) but also non-persistent: people had to go through the same process to revisit the site. This has now been fixed, but this was a poor user experience that should never have happened. Moreover, the MPF system has never been popular in Hong Kong, and this initial failure reinforces the sense that retirement savings are remote and not truly member-centric.

MPFA acknowledges these problems, citing challenges of adapting to a new operating environment, the continued use of old trustee reference numbers, and the cleansing and standardization of member data from different trustees.​

The eMPF Company says it responded with more manpower, longer hotline and service-centre hours, a contribution inquiry hotline, dedicated complaint officers, and closer contractor oversight.​ However, Lui did not respond to questions specifically about the rollout or PCCW’s performance or accountability.

Nor did MPFA respond to our request to disclose metrics for active users, defined by those who have used the platform more than once. Active usage is a better benchmark of whether members value or trust the eMPF platform enough to make it a regular part of their financial lives.

The official vision and its gaps

Lui said eMPF “paves the way for future digital reform initiatives” and that the eMPF Company will work with the MPF sector and the wider finance industry to explore collaboration and data sharing, where permitted by law, to enhance user experience and offer more integrated financial services.​

This comment should be welcomed. It recognizes eMPF as more than an administrative cost-saving tool and sees potential for it to become part of Hong Kong’s broader financial infrastructure.​

Still, the official framing remains cautious and incomplete. The primary policy objective, as stated by MPFA, is still cost savings and administrative efficiency.Those goals are important, but they are not sufficient.

The true promise of eMPF lies in using this centralized architecture to enable open APIs, data portability, independent digital advice, fuller fee transparency, and eventually greater portability and competition among providers.​

On these questions, MPFA’s response is an acknowledgement, not a roadmap. It signals openness to collaboration and enhancement, but offers no concrete policy agenda for APIs, no timetable for portability or consent standards, no framework for third-party advice, and no indication that member-directed market competition is yet a strategic priority.

Why openness matters

This matters because MPF is not a minor savings pool. It is one of Hong Kong’s largest long-term domestic capital bases, and one that sits largely outside the day-to-day financial awareness of most members.​

One of its handicaps is structural, beyond the scope of digitalization: the system is mandatory but the minimum monthly contributions are small, and capped at a low base (although members can supplement this with voluntary contributions). In aggregate, MPF assets are large, but they play only an ancillary role in most people’s retirement finances. Launched in 2002, the system is young compared to those in Singapore, Australia, Japan, or Great Britain, both in absolute and in per capita terms. Kept within its own silos (in terms of investments, administration, and regulation), MPF matters but it doesn’t matter enough.

If eMPF remains chiefly a centralized back-office utility, it may succeed on administrative terms while falling short of its potential to improve retirement outcomes, stimulate product innovation, and connect retirement savings more intelligently to the broader financial system.​

An open-finance approach would not mean abandoning regulation or member protection. On the contrary, it would require carefully governed access, consent-based data sharing and clearly licensed participation by trusted third parties. But if designed well, it could let members view MPF alongside bank savings, insurance, mortgages and investments; receive more tailored retirement planning; compare fees and products more clearly; and use specialized digital tools developed by fintech firms rather than relying only on the narrow interfaces of incumbent providers.

MPFA’s own response hints at this logic when it says trustees can redirect resources toward higher-value functions such as investment management, analysis and advisory services.​ The natural next question is whether those higher-value functions will remain largely internal to incumbents, or whether eMPF will evolve into a platform on which a broader set of authorized firms, including technology companies, can compete to serve members better.

In short, open finance would leverage MPF into something that matters a lot more for Hongkongers, and for the financial services industry.

An ambitious way forward

The strongest version of this argument is that Hong Kong has already done the hard and important work of building the core rails. That deserves recognition. Yet building the rails should create the confidence to ask what should run on them next.

An agenda combining practicable steps and a medium-term vision would include four priorities.

• First, stabilize and improve the user experience, with more transparent reporting on complaints, repeat usage, onboarding success and service-resolution times.​ The eMPF Platform Company has already gone some distance in ameliorating the onboarding process, but this could be enhanced, and a better marketing effort may be required.

• Second, publish a clearer long-term policy vision for eMPF as financial infrastructure, not just as administrative reform. This is traditionally a challenge for MPFA, whose culture is one of administration, not policy nor taking career risk. Any substantial measures must come from higher up in government, at the level of the Financial Services and Treasury Board, chaired by the finance secretary of the Hong Kong government. But MPFA houses the expertise. Other financial regulators such as the Hong Kong Monetary Authority and the Securities and Futures Commission have in recent years adopted a more proactive stance toward promoting fintech and digitalization, including market promotion and legislative agendas in areas including blockchain and artificial intelligence. It is time for MPFA to do the same, in dialogue with other regulators, lawmakers, and market participants.

• Third, revisit portability, fee unbundling, independent advice, and participation of vetted technology companies, so the system becomes more genuinely member-driven over time.​ These are within MPFA’s remit to propose. The previous structure of decentralized administration mitigated against such competitive moves, and reform was incremental and insufficient. Over time, if user engagement grows, there will be a grassroots demand for a more competitive and responsive MPF system. MPFA has the opportunity to prepare now and articulate what may be possible in the medium term. Notably this would set the stage for open finance.

• Fourth, begin formal work on open API and consent frameworks that would allow secure participation by banks, insurers, advisers and fintechs. This is a longer-term goal but one that can be articulated today. Hong Kong has a broader challenge when it comes to open finance, and there is little the MPFA could do without deeper changes to how Hong Kong legislates and governs open finance. However, a show of interest by MPFA could represent new business opportunities for financial institutions to build use cases for personal financial management. MPFA may not be able to act alone on building open-finance rules, but it can stimulate discussion within the broader community, which would be welcome even beyond pensions-related opportunities.

The MPFA has shown it is willing to engage in at least parts of this conversation, even if it has not fully embraced the implications of the eMPF Platform and putting this information in the hands of its users. This article argues for the MPFA to assume new initiatives, but it is also a call for market participants to encourage the MPFA and the FSTB to take these proposals seriously.

With the eMPF onboarding complete, it is time to start a new dialogue that moves toward a more expansive public vision: one in which eMPF lowers costs, wins trust, broadens access to better advice and opens a major pool of long-term savings to wider forms of innovation and value creation within Hong Kong’s financial system.

There are some good reasons why ESG is politically bankrupt in the US. The “G” of governance is already an established aspect of financial and business analysis. “S”, social factors, is too fuzzy a concept, one that changes radically from one country to the next, to be neatly calculable. “E” is critical to human flourishing, but it’s more than just about carbon, and this complexity has led to too many solutions and not enough standards.

ESG was invented by management consultants, so it’s not a surprise that it’s become so complex as to be meaningless.

But the physical risks to businesses are still there, and the lights are flashing red. Moreover, US politics aside, boards and investors worldwide are beholden to ESG reporting requirements, and in Europe and China, regulation around biodiversity and carbon footprints have teeth. And everywhere, such exposures are being baked into infrastructure and energy policies through the lens of national security.

Dunya Analytics is part of a new generation of ESG-related tech companies that are using data to help CFOs and risk managers understand where their exposures lie when it comes to nature. Its founder and CEO, Megan Pillsbury, has served in senior tech and AI roles at Goldman Sachs and Morgan Stanley, in Asia and the US. She is now based in Wilmington, Delaware.

Timecodes:

0:00 - Megan Pillsbury, Dunya Analytics

2:36 - Political divergence in ESG

4:42 - Lessons from on-chain voluntary carbon-credit markets for the nature space

9:21 - Regulation, incentives, and tech in building new markets

12:17 - Helping companies and investors measure their exposure to nature; and the challenges of many measurements and too few benchmarks

18:39 - AI and where it’s valuable and where it’s not

20:56 - Trends in fintech in nature and carbon, and where globally to find the most action

25:07 - Stranded assets in biodiversity and carbon

Last week, mid-April, the payments and fintech conference group Money20/20 staged its annual Asia leg, in Bangkok. This was the organizer’s third year there, and Thai banks mostly ignored it: Kasikornbank was a minor sponsor, but Bangkok Bank, Siam Commercial Bank, Krung Thai Bank, and other lenders were a no-show.

This is more a loss for Thai banks than for Money20/20. It’s a testament to the inward focus of Thailand’s financial institutions, which has hampered the flourishing of Thailand’s fintech industry. Might change be coming?

Any national bank should be focused on its domestic business. But Thailand is part of the Association of Southeast Asian Nations, whose members continue to build infrastructure to connect their financial systems. Its central bank was a member of mBridge, a BIS-blessed platform for exchanging central-bank digital currencies, with China, Hong Kong, and the UAE. More than 1 million Thais work abroad, remitting money home. Bangkok is a travel hub between Europe, the Middle East, and East Asia.

Having a global finance and fintech conference hosted in Thailand’s capital would have given local banks exposure to new ideas and companies. Event fees weren’t an issue: the top banks are highly profitable, and the conference was discounted compared to its US and European sleeves.

The snub was because Thai banks feel comfortably ensconced. They have been good at fending off startup competition. Although they have their own digital agendas and venture investment arms, local banks deploy these to ring-fence the status quo: the imminent launch of three digital-only banks is another iteration of that strategy. The presence of many stablecoin-related companies at Money20/20 was not a draw; possibly the opposite.

Although Thailand’s leading banks are adept at serving retail customers, they continue to mostly ignore SMEs. They are weak at cross-border business. These gaps have not overly bothered the banks. But the possible arrival of open-banking frameworks could shake things up, by making SMEs and their cross-border payments needs a meaningful opportunity for startups and new entrants.

Fintech false dawn

Thailand’s fintech sector no longer looks like the easy growth story it once promised.

Its history goes back to 2003, with the launch of TrueMoney, later restructured as Ascend Group, which runs the country’s leading mobile wallet; and of 2C2P, an online payment provider that later moved to Singapore (and has since been acquired).

The mid-2010s made fintech a trend, with the debut of companies including payment gateway Omise (today, Opn), cross-border payments player DeeMoney, wealthtechs Finnomena and Robowealth, and insurtech Roojai.

They are the survivors of a wave that included dozens of startups. Indeed, today the country claims up to 177 active fintechs (depending on who’s counting), led by payments and alternative lending. But those numbers are surely misleading.

“Eighty percent of Thai fintechs are zombies,” said Kris Supavatanakul, strategy director at Finnomena and a former corporate VC at Siam Commercial Bank.

One reason is the loss of venture-capital funding, a global trend. At the height of the zero-interest-rate funding boom, in 2021, Thai fintech firms raised $216 million. That number obscures the fact that most of it involved Ascend Money’s monster $105 million Series C round. Fintech funding has fallen off a cliff since, both in Thailand and across Southeast Asia.

Bank dependent

There are more fundamental challenges to Thai fintech. In most areas of finance, fintechs depend on commercial banks for infrastructure, technology, and distribution.

Payment infrastructure is monopolized by National ITMX Company, established by the Thai Bankers’ Association under the direction of the Bank of Thailand’s payments division. ITMX is responsible for developing and maintaining the electronic payment infrastructure, including mobile, internet, ATM, and other rails, such as the QR system PromptPay.

“The national switch serves the banks,” said Aswin Phlaphongphanich, co-founder and CEO of DeeMoney. “That means fintechs can’t be agnostic. We don’t have technological independence, which means we don’t have commercial independence.”

The problem for fintechs is simple. The industry is almost entirely focused on retail customers. To serve them requires using ITMX’s switch, operated by the commercial banks (at arm’s length). In theory, a fintech could build its own payments infrastructure. But four banks control more than 80 percent of retail deposits. They are also on top of digital offerings.

Things get even more complicated for fintechs. The majority choose to simply serve banking partners with user-interface design, KYC services, or other operational services. But if a fintech wants to touch customer money – to collect, disburse, or settle – they are locked out. They can participate on ITMX rails if they are licensed, but these are hard to get. Nor are all licenses equal: often, licenses granted to fintechs are restricted, and come with the sort of red tape that big banks can absorb but that startups cannot.

Finally, in the mid-2010s, banks began eliminating fees on mobile payments, back when few people saw these taking off in Thailand; today, mobile wallets are a big business – for retail users. They don’t make any money for fintechs, and banks absorb the cost into deposit businesses.

Local heroes

There are, of course, fintech success stories in Thailand. They involve companies that have found a way to avoid relying on banks for moving money or access to users.

• Ascend Money is owned by CP Group, a massive conglomerate whose empire includes the domestic 7-11 franchise. The fintech has more than 20 million users and it can charge a fee for facilitating consumer purchases from its retail affiliates.

• Finnomena has built its own brokerage business to sell mutual funds. It continues to build this out, and is seeking a full brokerage license, so it can also execute orders for individual stocks and bonds. But it does not use a bank channel for acquiring customers.

• DeeMoney has found a niche in cross-border payments, because Thai banks ignore this space. Omise specializes in online payments for businesses in the digital economy, because banks ignore SMEs.

A few foreign fintechs have also succeeded in Thailand.

Funding Societies, based in Singapore, has become the biggest fintech lender to SMEs. It entered in 2021 as one of the first fintechs to receive a debt crowdfunding license from the Thai Securities and Exchange Commission. Funding Societies already had a thriving business in regional markets, matching SME borrowers and investors looking for yield. Its story shows what happens when a competent company that spends years on obtaining a license is able to serve credit-starved SMEs.

Singapore-based insurtech Igloo has announced a joint venture with local telco JMT Network Services to launch Thailand’s first fully digital insurer. Early days!

Wise Platform is about to launch in Thailand. After many years of assiduous courting, in March the remittance platform secured five licenses (for payments, electronic money, and FX) from Bank of Thailand and the Ministry of Commerce. This makes Wise the first non-bank in the country to be fully licensed to issue foreign-currency wallets and cards. But the licensing was only one part of London-based Wise’s breakthrough. Its global remittance business relies on having a bank account in the local market, from which it transfers customer debits and credits. One of the Thai commercial banks broke ranks and has agreed to provide a deposit account to Wise; its identity has not been announced.

Enter the virtual banks

One factor that could explain this class betrayal is the advent of three licensed digital banks, which are slated to go live this summer. These consortia are more likely to cannibalize the commercial-banking market than innovate any new products or services.

Two of them are partly owned by big local banks. Siam Commercial Bank is partnered with South Korea’s KakaoBank and China’s WeBank. Krungthai Bank has a more domestic consortium, with telco AIS and PTT Oil & Retail. The third group includes CP Group’s Ascend Money.

These three virtual banks are all focused on the retail population. They bring capital, data, and distribution. They do not include disrupters or banks with wholesale arms. There is no equivalent to, say, Grab’s leading position in Singapore’s retail digital bank, GBX, or an e-commerce outsider like Shopee (a stakeholder in Mari Bank).

Indeed, a look at Singapore shows that its retail-focused digital banks face major obstacles to growth. The population is already banked. Trust Bank (backed by Standard Chartered) has a huge retail footprint, thanks to its shareholder, NTUC, but it can’t make a profit. The two wholesale players, ANEXT Bank (Ant Group) and Green Link Digital Bank focus on SME trade and supply-chain finance, and are flourishing.

There doesn’t appear to be an equivalent story among Thailand’s new trio of digital banks. The consortia appear designed to protect incumbents, not to unleash new categories of competition.

This could bring risks to the banking industry. Thailand’s population is fully banked. It’s so well banked, that it also has a huge household debt problem. With a debt-to-income ratio of 88 percent (as of Q3 2025), Thailand’s consumers are the most indebted in ASEAN. Virtual banks tend to make headway by offering higher interest on deposits combined with slick user interfaces. What’s likely to happen is a huge refinancing of debt, which will move the problem out of traditional banks and onto the books of the three virtual banks. This is a recipe for rising non-performing loans, which will eat their capital base, obstructing rather than enabling them to innovate with fintech partners.

Although the three new banks could provide some new distribution or offer infrastructure, they are unlikely to help Thailand’s fintech industry to grow. Most fintechs focus on retail, for payments, wallets, lending, and investing. As we’ve seen, only a handful of them control their own destiny by avoiding depending on banks for distribution, tech, and licensing. As a class, it is hard to see Thai fintech scaling beyond the nation’s borders.

Open doors

The key to creating new opportunity for fintechs is not virtual banks, but open banking.

And there is good news on this front. The Bank of Thailand has declared an open data framework gives customers the right to port data across deposits, loans and payment services through secure and standardized digital channels. This is meant to phase in at the end of 2026, initially for individual deposit data. Later, business data and other categories are meant to go live.

If implemented seriously, the BoT’s policy would begin to loosen one of the deepest constraints in Thai fintech: dependence on banks not only for settlement, but also for customer data, onboarding pathways, and product distribution.

The promise is clearest in lending and SME services. Open banking could allow fintechs to refinance debt, consolidate deposits and build better borrowing journeys once data and rails are truly accessible.​ The Bank of Thailand has also explicitly tied digital finance policy to household and SME financial inclusion, while acknowledging that SME access to credit remains structurally weak.

But policy intent is not the same as implementation. Thailand’s open banking push could still devolve into a compliance exercise if APIs are technically available but commercially weak, or if major banks delay, limit or shape access in ways that protect their franchise. The framework only changes the market if a few big banks actually lead, rather than stall.​ Given how concentrated Thai banking remains, skepticism is warranted.

The SME opportunity

The strongest argument for a second wind in Thai fintech is not another retail super-app. It is SMEs. SMEs account for nearly all Thai enterprises, yet only about half of them can access formal credit from banks or specialized financial institutions, and SME loan growth has been weak or negative.

Existing fintechs and banks alike still skew toward consumer products because retail is easier to score, market and distribute at scale. With open banking, the incentives to tackle SMEs become too great to ignore. A few fintechs such as DeeMoney already operate an API-based platform for global remittance companies such as Western Union and Remitly to disburse funds into Thailand without touching a commercial bank branch.

There are plenty of new business ideas: collecting baht for multinationals who want to repatriate money home; stablecoin settlements; global QR-based settlements. These don’t require a bank partner, but they do require a license or regulatory green lights. Open banking would make it hard for regulators to say no.

Thai fintechs are most competitive when they solve coordination problems that banks don’t prioritize: cross-border remittances, SME collections, overseas e-commerce settlement, regional treasury flows, fragmented ASEAN payment corridors. (The exception is Bangkok Bank, the biggest Thai lender, which also operates more than 30 branches overseas, mostly in Southeast Asia and China, and a controlling stake in Indonesia’s PermataBank. About 25 percent of its loan book is sourced internationally.)

Thailand’s domestic economy is not large enough to support endless new consumer-fintech clones, but Thailand as a node in ASEAN trade, labor mobility and cross-border commerce is much more promising. That is particularly true for SMEs that import, export, sell online or rely on regional supplier networks yet still struggle with FX pricing, slow settlement, poor visibility and bank-centric onboarding.

Because licensing is so slow and painful for fintechs, they need a bank shareholder or partner to survive. Open banking as real infrastructure, not a regulatory slogan, would pave the way for standardized APIs and consent frameworks. This would pressure banks to make data and payments usable for third parties. It would create incentives to allow fintechs and banks to pivot toward SME pain points. And it would put pressure on regulators to create licenses that allow non-banks to scale on their own.

Cross-border and SME services are where innovation is needed, and can happen. It’s notable, therefore, that a leading global conference that’s all about financial connectivity and has taken place in Bangkok for three years is shunned by Thai banks. It’s not a conversation they want to have.

But open initiatives often stall, or get bogged down in costly compliance or tech requirements. These barriers derive from the source of open finance: regulation.

The United Arab Emirates is taking a new approach. The Central Bank of the UAE has set up a new, commercial business called Nebras, to serve as an API hub, providing standardized connectivity. Financial institutions of a certain size are mandated to plug in across banking, insurance, and FX.

Jonathan Holman has been tasked with bringing Nebras to life. As the CBUAE’s head of open finance, he has been named chief executive of Nebras. His background includes academia, consulting, and three years as head of digital for SME, commercial, and corporate banking at Santander in the UK.

Timecodes:

0:00 - Jonathan Holman, Nebras - now incorporating an ecosystem entity running national infrastructure for open finance

4:58 - A federal initiative for fintech and data consumption and interaction, with API connectivity as a condition of financial licensing

7:41 - The pros and cons of building a centralized infrastructure for APIs

11:15 - Who is obliged to participate in Nebras - banks, insurers, fintechs, wallets, third-party users of data?

13:47 - Nebras’ applicability within the UAE’s patchwork of regulatory zones

16:03 - Use cases in open finance and how “open banking” has evolved

18:27 - The ambition for open finance in the UAE

21:13 - AI and agentic capabilities in open finance

23:59 - Liability in agentic use

26:23 - A single point of failure?

28:01 - How Nebras fits alongside blockchain rails

29:46 - Prospect for cross-border open finance

Is DeFi dead? One of the leading lending platforms of decentralized finance, Aave, has been torpedoed by a de-facto run. It is one of multiple cybersecurity failures hitting the industry. Dead or reborn is not fated but a choice, by the many people in blockchain finance. But the time of casual disregard for fusty TradFi protections is over.

As a lending protocol, Aave is kind of like a giant crypto bank, where anyone can deposit coins and earn interest, and anyone can also borrow coins if they put up enough collateral. It’s all run by code instead of bankers, and until recently it was seen as one of the safest places in DeFi.

Now imagine a special kind of crypto token – call it “staked ETH points” – that’s supposed to be backed 1:1 by real ether locked somewhere safe. A separate project (KelpDAO) issued those points and used another piece of infrastructure (LayerZero, a type of bridge that moves tokens between blockchains) to move them around. The important bit: everyone treated those points as solid, reliable collateral, including Aave.

Then someone found a way to break that setup. On Saturday, April 18, they managed to pull out a huge amount of those “staked ETH points” without proper backing, and then went to crypto lending platforms like Aave and said: “Here’s my collateral, please lend me real ETH and stablecoins against it.” The platforms couldn’t tell the difference between good and bad tokens, so they handed over hundreds of millions of dollars in real assets.

By the time the problem was noticed and the bad token was blocked, it was too late. The attacker had borrowed real money and disappeared. Perhaps about $290 million’s worth was stolen. What remained inside Aave and other platforms was a hole: lots of fake or worthless collateral, perhaps as much as $230 million’s equivalent, and a big chunk of loans that will never be repaid.

Run run run

On paper, Aave still has more assets than debts. In theory, that means it’s not bankrupt. But that is not what users experience.

Because of the exploit and the panic it triggered, a lot of big players rushed to withdraw their money from Aave. That drained the pools of assets that everyone shares. For many key coins – especially ether and major stablecoins like USDT (Tether) and USDC (Circle) – almost all the money in the pool ended up being lent out, and almost no spare liquidity was left.

If you deposit money into a pooled lending system, your ability to withdraw depends on some cash being left in the pool. When that “available cash” goes close to zero, you simply cannot get your money out, even if you’re technically entitled to it. The code won’t let you withdraw what isn’t there.

That is exactly what happened. For some markets on Aave, utilization – the share of deposits that are lent out – shot up to nearly 100%. (Unlike in traditional banking, there is no reserve requirement.) Users opened the app and found they couldn’t withdraw. Some tried a messy workaround: borrowing different coins against their stuck deposits, then selling those new coins elsewhere at a loss just to escape. In other words, they accepting taking a haircut.

To make matters worse, people who had borrowed stablecoins against ether as collateral suddenly could not repay their loans normally either. The markets where their collateral sat were frozen: about $5 billion worth of USDT and USDC are, for now, frozen, like deposits that a bank won’t redeem.

If ether’s price had crashed during this period, Aave’s systems would have struggled to liquidate positions, which could have created even more bad debt. Fortunately that hasn’t happened.

Nonetheless, from the outside, what has happened on Aave looks and feels a lot like a bank run: the first to rush out get paid, everyone else is stuck. Users have pulled more than $6 billion of ETH-denominated positions from the platform since news of the exploit got out. Aave’s TVL, total value locked, fell from about $26 billion to $20 billion.

Three problems

While industry insiders are still investigating the details of what’s happened, three big problems are obvious.

First, as we’ve said before, it’s the plumbing that matters. But in DeFi, the plumbing is treated as an afterthought. Most people, and most risk models, focus on the visible token: how volatile it is, how much you can safely borrow against it, what price feeds are used. But the real risk was hidden below: who runs the bridge that moves the token between chains, who holds the keys, how that system can fail, and what happens if it does.

When Aave accepted this “staked ETH” token as collateral, it wasn’t just trusting the token. It was trusting an entire stack of software and operations it didn’t control or deeply understand. That’s like a bank accepting a new type of mortgage asset without really examining the company that created it, the custodian that holds the paperwork, and the system that tracks ownership. Remember subprime mortgages? The evils of TradFi that DeFi was supposed to transcend?

Second, shared pools are transmission engines. Aave and similar platforms use shared pools: everyone deposits into one big pot, everyone borrows from the same pot, and interest rates adjust automatically. That’s efficient and convenient. It also means that when a single type of collateral blows up, its effects spread everywhere.

In this case, a problem with one token tied to ether ended up freezing markets in ether itself and even major stablecoins. People who had never heard of KelpDAO or LayerZero suddenly found they couldn’t move their USDT or USDC in Aave. Their only mistake was using the same pool as everyone else. Ah, remember how Lehman didn’t separate customer assets under its prime brokerage business, and we all had to learn how to pronounce “rehypothecation”?

Third, who’s on the hook? When a big hole appears, someone has to pay to fill it. In traditional finance, we have a whole hierarchy: shareholders, junior bondholders, senior creditors, deposit insurance, government backstops.

In Aave’s case, there is a sort of insurance system called Umbrella. People stake ether into a pool that can be slashed (cut) to cover bad loans, in exchange for rewards. The problem is that Umbrella was designed for relatively ordinary problems – like sudden price swings and liquidation cascades – not for a single event that might wipe out hundreds of millions of dollars linked to a bridge bug or misconfiguration.

So now the system is in a bind. If the loss is larger than this insurance pool, the platform’s community treasury might have to step in. If even that isn’t enough, some of the loss may have to be pushed onto ordinary depositors. The rules for who gets hurt, and by how much, are being figured out on the fly. The DeFi industry must use this disaster to establish, uh, rules? Enforceable by law, not just code?

Vercel and centralized compute

Governance is not just specific to bank-like runs in DeFi. Cybersecurity problems are plaguing all of DeFi. In this regard, crypto isn’t difference from TradFi, where the same risks are manifest. But crypto is, well, crypto native, which makes the industry even more vulnerable.

While the Aave mayhem was going down, a separate incident occurred. Vercel, a popular cloud service that many crypto projects use to host their websites, reported a security breach. Attackers may have accessed internal systems, including keys and deployment credentials for customer projects.

What this means: even if the smart contracts running your money are safe, the website you use to interact with them could be compromised. A hacker who controls a project’s frontend can trick you into signing malicious transactions that drain your wallet, all while showing you a normal‑looking interface.

What else this means: many “decentralized” projects still rely heavily on centralized, traditional tech companies for critical parts of their stack. So even if the blockchain is resistant to censorship or tampering, the path you use to reach it might not be.

For ordinary users, that’s yet another reason to feel that DeFi is unsafe: hacks are coming not just from clever contract exploits, but from weak links in hosting, software supply chains, and development tools.

Sink or swim

Again, not limited to crypto, but crypto is more vulnerable. Specifically, the DeFi story is no longer viable. Putting finance on-chain doesn’t make it safer. Clever software engineering plus a few audits is not enough to handle serious amounts of money. Cutting out bureaucracy and oversight, without replacing them with equally robust software, is efficiency-maxxing – for criminals.

Traditional finance has become disparaged as “TradFi” because it too experienced this kind of blowup. And more. Controls, capital buffers, governance, and regulation didn’t materialize because anyone asked for them: they came after many painful lessons. DeFi’s undergoing a similar education, but a lot faster, and visibly, as the transactions occur as hashes visible to everyone on-chain. That transparency is a feature of DeFi, one of its best attributes. But code alone isn’t protecting the marketplace. Code is not law; it’s just code.

Once the dust settles, how does DeFi recover?

• First, we may see a winnowing of the kind of tokens that circulate. People will trust plain ether, major stablecoins, and well-scrutinized tokens. Others will be eyed as a risk. Bridged tokens, complex derivatives, and many restaking products will be tagged as risky; they won’t vanish but they will be restricted to the shark tank, not the pool for the general public.

• Teams have to undergo root and branch reform. It’s time to get Old Testament on those teams that laugh at compliance culture or think speed is all you need.

• Insurance and other backstops need to be sized to match the level of flows transacting across a platform. TradFi bankers always chafe against capital reserve rules, so this isn’t a dynamic unique to crypto, but the pendulum in DeFi has to swing a little harder.

• Institutionalization in this space has been on the rise for the past two years, enabled by new regulation and licensing regimes. A lot of that activity is going to DLT rather than DeFi, to protocols such as Canton Network that allow like-minded institutions to transact among each other, with no riff-raff allowed. Institutions can benefit from DeFi’s greater reach and liquidity, but they need confidence to venture here.

DeFi needs to work with institutions to learn what they can without sacrificing the promise of permissionless, decentralized, efficient capital markets. There are plenty of traditional brokers, market makers, and traders who want to play in DeFi-like environments: the shark tank is fine, so long as everyone knows what’s swimming in there, and the sharks can’t jump into other pools.

These things will take time, the one thing that DeFi doesn’t have. The immediate issue is restoring liquidity so user stablecoins on deposit can be withdrawn. Aave will have to find a way to convince users not to pull out, but it has to restore their ability to do so. Beyond one platform, DeFi should rethink the story about itself. Otherwise it will be remembered as just another speculative bubble.

Maximilian Plank is a venture partner at Tenity, a VC and accelerator program focused on fintech, based out of Zurich.

Max’s career has involved helping financial centers interact. He serves as a ‘global ambassador’ to the Middle East for Hong Kong’s Cyberport, a government-supported startup hub. He chairs a business facilitation startup in Doha, and serves as Tenity’s lead partner for Saudi Arabia.

He speaks with Jame DiBiasio about the current hot themes in fintech, globally and in select markets, including sharing some of his investment companies, the opportunity in digital assets and in AI, and the impact of the ongoing war in the Gulf on investing.

Timecodes:

0:00 - Max Plank and Tenity

2:50 - What do LPs look for in fintech investments today? The difference between venture investing and startup acceleration

6:00 - Fund composition and a new strategy for digital-asset infrastructure

9:00 - Fragilities in blockchain, startups with a license or not, and new business advantages in the age of AI

15:03 - Analyzing founders in the new environment

17:12 - Switzerland: old banks and new foundations

21:38 - Saudi Arabia: war, fintech, and opportunity

27:29 - Hong Kong: tokenization, links to the Middle East, and a new generation of Chinese fintechs

The usual answer in finance to an either-or question is ‘both’. That’s true of tokenization, where the action is gathering around two protocols, one that is designed for permissioned, club-like activity, and another that is public and open to all participants. Depending on the objective, a bank or asset manager can operate on Canton Network for a closed-door environment, or on Ethereum if they want to reach a far greater universe.

At some point, though, ‘both’ becomes hard to achieve.

Today more than $29 billion of real-world assets are deployed on-chain, according to rwa.xyz, with more than $15 billion of that distributed through Ethereum. That distribution represents asset value of $354 billion, of which $303 billion is stablecoins.

Ethereum’s position represents stock: the value of RWA represented on-chain. But when measuring flow, the cumulative notional value of transactions – use of collateral, secondary-trading volume, total settlement value – the clear winner now is Canton, which processed more than $6 trillion in RWA in 2025, mainly in repo and other wholesale flows.

This reflects how banks and asset managers are using these platforms.

Bifurcation

Ethereum and its competitors (Solana, a direct competitor, plus other chains that interoperate with Ethereum such as Avalanche, BNB Chain, and Polygon) provide transparency, reach, and composability of the token. They are important for distribution, investor access, and secondary-market liquidity.

Canton Network – and other permissioned distributed-ledger stacks such as Corda, Hyperledger Fabric and Enterprise Quorum – provide bilateral trust, regulatory comfort, and confidentiality. These are where banks go for dealer-to-dealer settlement, intraday collateral mobility, and internal treasury flows. Canton’s ‘killer app’ is processing Broadridge’s DLR, a platform for repo and securities financing among buy and sell sides.

In sum, one rail brings chaos, but the other brings opacity.

This bifurcation raises two questions for financial institutions. The first is that each rail by itself brings its own constraints: legal, operational, market structure. What happens when an institution hits a buffer?

The second question is how institutions will address the fragmentation of liquidity this brings. TradFi doesn’t have this division of activity into such distinct buckets. For now, tokenization markets are nascent. How big do they have to get before this divide becomes a constraint? How expensive will it be to manage separate liquidity pools?

It’s possible that this fragmentation problem will gradually solve itself. But today, moving assets between chains is expensive, adding up to 5 percent in cost, and creating pricing gaps between venues, which for now, arbitrage is unable to address.

Blockchain is touted as being efficient because traditional assets become cheaper and faster to issue, move, and pledge. Static holdings such as T-bills or money-market funds become programmable collateral. Composability makes it easy to offer such assets to new investors without reinventing compliance.

But at scale, this either/or problem would undermine any such efficiency gains. Do the Canton and Ethereum worlds converge – or harden into separate worlds?

Ethereum

That may depend on the flexibility of the contending platforms.

By the end of last year, regulators in major financial centers had turned Ethereum’s token an institutionally acceptable asset, thus making Ethereum the default public venue for tokenized RWAs.

Most use cases today involve tokenized funds and T-bills, stablecoin payments, and a few funky capital-markets products such as private credit and structured products. Every asset, every position, every transaction is globally auditable and visible to the network. Any wallet, custodian, or fintech can participate, negating new distribution costs. Tokenized assets are composable, meaning they can plug into lending markets, automated market makers, and other aspects of DeFi.

All of this has made institutions cautious about Ethereum and its ilk. DeFi means operational risk. The industry is rife with hacks and failed bridges (layer-2 chains). It’s still amateur hour at many of these projects.

Even with improved L2s and privacy tools such as zero-knowledge proofs, institutions can still end up revealing too much about their positions and flows.

And who do you call when le merde hits the fan? Ethereum is ‘run’ by a decentralized foundation.

Moreover its technical founder, Vitalik Buterin, is campaigning for the network to return to its cypherpunk roots, meaning its developers may lean harder into censorship resistance, privacy, and neutrality.

This would suggest Ethereum could become more like a public infrastructure that cannot be customized to enterprise preferences (such as whitelisting or access controls). This could also align with institutional interests, making it more auditable, better governed, more trustworthy, and inhospitable to the ‘number-go-up’ crowd. With the likes of Solana biting at its heels, such changes would probably be positive for financial institutions, but it would come with tradeoffs. And the uncertainty means institutions, while craving Ethereum’s public distribution and liquidity, will hesitate before putting their balance sheet on it.

Canton

This is why many leading institutions are piling into Canton. It takes away a lot of the problems of Ethereum. For wholesale markets, a club is better than a public market. It enables institutions to move tokenized collateral among trading books in privacy; and RWA tokenization stacks for issuance and settlement remain inside a bank’s private subnet, shielded from scrutiny. Activities such as repo can scale inside this shell, and avoid all the messiness of DeFi and engaging with smart contracts from anonymous parties.

But this comes at a price. The state of a token is not anchored to a public ledger via ZK proofs, which means no one can independently verify asset supplies, flows, or aggregate exposures. They only see what their counterparties choose to disclose.

One of the theoretical benefits of tokenization would be risk management: more visibility would make it difficult for an Archegos-type situation to develop. Archegos was a hedge fund that borrowed massively from multiple banks against the same collateral, and when it went under, it cost its prime brokers dearly. But in Canton, there isn’t the kind of transparency to mitigate this kind of counterparty risk.

One of the benefits of Canton is it’s a club. But a club means internal ‘super validators’ and other influential functions are controlled by a committee. Today this gives banks a sense of comfort. But over time can also give incumbents leverage over who else gets to join the club, potentially blocking new competitors.

And then there’s reach. Assets native to Canton can’t readily be used in DeFi, or held in the wallets used by smaller institutions or retail investors. A Canton participant can syndicate the economics off-chain, creating synthetic exposures, but this brings back the inefficiencies that tokenization is meant to smooth over.

For now, those problems are theoretical. The big banks, asset managers, and card processors are happy to be in their own club for wholesale instruments. But at some point, competitive pressures will make them want some of their tokenized assets to be discoverable, tradable, and collateralized in the broader on-chain world.

Come together?

Nobody wants an either/or. They want both. The rails themselves may evolve to address their limitations.

Canton could increase external, independent trust without sacrificing the privacy that institutions like. Perhaps it could allow cryptographic tools to provide aggregated data to public chains, or introduced tiered memberships that layer in degrees of technical participants. It could overhaul the rights to Canton’s native token. (One of the distinguishing features of Canton Network is that, while it’s a permissioned club, its governance token trades on public crypto exchanges.)

We’ve already noted the debate within Ethereum’s developer community about its direction. Its question is how to keep its robust verifiability while improving privacy.

Some of this is about operating standards for DeFi players, such as around security procedures, but this could go further, toward introducing legal concepts around on-chain fiduciaries. Ethererum could also take the lead on making ZK-based privacy pools make the jump from cool ideas to audited, regulator-understood products.

Its biggest challenge may be its own messaging. Vitalik Buterin doesn’t ‘run’ Ethereum, he’s just one influential voice among a coterie of developers. But the foundation may need to work on how to communicate with institutions (and their risk and audit committees) – not to centralize decisions, but to explain what is going on.

Institutions, meanwhile, will need to think about their own approach. They can’t assume these protocols will converge enough on their own, or that liquidity will cohere. But banks and asset managers that create their own bridges? That would be the ultimate advantage in tokenized assets.

These are classic black-swan events, that few take seriously but which have devastating ramifications should they come about.

Central banks and regulators of digital infrastructure are debating ways to address the payments aspect of a potential, comprehensive loss of electricity. Because even if the lights go out, there can be ways to enable some degree of economic activity to continue.

The old-fashioned way is to use physical cash. It’s a good idea to keep a stash around! But this is the digital age. We have moved too much of our commerce and daily needs online to think that some paper banknotes are the only or optimal solution.

Debating this with Jame DiBiasio is Joachim Samuelsson, CEO of Crunchfish, a Malmö-based startup that has its own solutions to enabling digital money to work offline. He served as its chairman when the company was established in 2012, and has founded or led multiple tech- and security businesses in Sweden.

Timecodes:

0:00 - Joachim Samuelsson, Crunchfish

1:38 - How big a problem do outages present in real life? Building for resilience.

4:04 - Is redundancy the best way to manage risk, rather than building bespoke solutions? The need for better solutions.

7:35 - What does ‘digital cash offline’ mean? Why not just rely on banknotes? Different approaches by central banks.

10:00 - Making physical cash work digitally, versus making digital cash work offline; questions of hardware security (mobiles).

14:33 - New solutions that don’t bring their own systemic risks.

18:54 - What Crunchfish is providing to India’s NPCI

21:09 - The challenge of getting lots of people to lock up money persistently. The impact of outages on individuals versus merchants, issues of scale.

28:30 - Central bank concerns: device security.

31:48 - Central bank concerns: double-spending and consumer confusion.

Financial institutions have always been in the business of warehousing risk: credit, duration, liquidity, operational risk. But guess what? It turns out the biggest risk on the balance sheet is the bank’s own software vulnerabilities; the platforms and protocols aren’t just revamping the exchange of value, but of attack vectors. The vaults and smart contracts aren’t just full of money and credit. They’re overflowing with bugs. You’re banking bugs.

This may not be news, but this week’s leap in AI means that inventory of software holes and glitches has just changed from being an IT/CISO to-do checklist to a balance-sheet event.

On Monday, April 7, Anthropic published its “Claude Mythos Preview”, unveiling details of its next-frontier large-language model, which will not be made public in its current form, yet. No matter, this is the current state of AI, and other labs, and not just American ones, will soon have similar capabilities.

The Mythos report (a ‘system card’ in Silicon Valley-speak) describes a model that can autonomously identify and develop exploits for previously unknown zero‑day vulnerabilities across every major operating system and browser Anthropic tested. And Mythos spots these in hours rather than the months it can take elite human red teams (in-house teams of ‘white hats’ who act like a real attacker). These include long‑standing bugs in hardened stacks that underpin critical infrastructure.

Finance’s fragility

For the financial sector, that matters because the same code is everywhere. Core banking systems, card processors, custodians, high‑frequency trading platforms, payments rails, cloud‑hosted fintech apps, DeFi nodes – they all sit atop this shared substrate of operating systems, compilers, crypto libraries, and virtualization layers. The sector has spent decades layering controls, processes, and capital buffers on top of that substrate without ever being able to – using a financial term – mark to market its fragility.

Mythos‑class models change that. They remove the human‑time bottleneck from vulnerability discovery. This is great for companies that can act quickly to patch things. It’s a disaster if you can’t. Anthropic says it’s quietly disclosed thousands of vulnerabilities to vendors. But that’s just the start.

Bank technologies have long understood their institutions house ‘technical debt’, the accumulated work and risk needed in software from choosing quicker, easier, and cheaper solutions rather than building robustly. Although the debt is high, the infrastructure has been sufficiently walled, gated, and moated to remain defensible. Now the economics of machine speed render the castle both suddenly vulnerable, but also allows defenders to modernize those firmaments.

But vulnerabilities that go unaddressed are now a liability, financially, legally, and by reputation.

Repricing risk

Sporadic samples of the risk book must now be treated as full-book, real-time marking. The book, in this case, being the entire software and infrastructure stack. Period penetration tests and compliance checklists are about to be repriced.

That repricing is going to happen first in DeFi. Smart contracts, bridges, and wallets are now honey pots. Immutable contracts with large TVL (total value locked) have been built without accounting for this kind of threat. The entire ecosystem reuses key aspects of data: smart contract builders, standard automated-market-making templates, oracle interfaces, layer-2 bridge designs, which could create a systemic risk to many protocols.

Relying on Big 4 audits won’t cut it now. Serious teams will move toward continuous AI-augmented auditing pipelines. If crypto trades 24/7, audits will also be non-stop and real-time. Market players will demand transparency around disclosing Mythos-class or equivalent models against a protocol’s stack.

The emphasis on shipping front-end product and UX (which is typical across TradFi and enterprise generally) will have to include a narrative around security, which implies a cultural shift. New products that are embedded with persistent, Mythos-level protections are going to be key to revenue growth.

These days digital assets have become institutionalized, and the bitcoin bros have learned to recite the letters ‘K-Y-C’, so this represents an acceleration rather than a deviation, but there remain plenty of slapdash teams, yield-farm casinos and meme protocols that will now be found out.

Quantum qualified

Crypto loves to talk about how AI is perfectly suited for its mission of agentic finance. Still true! But the industry is already a seeping cesspit of hacks and fraud (see: Drift). The next year is going to be wild. But crypto is just the thin edge of the financial wedge.

This will also transform the new fashion around worrying about quantum computing and its ability to smash cryptography. The reality of quantum computing is unclear, but it’s still a story told in years, if not decades. AI has shrunk that timeline to months.

Mythos doesn’t attack the math of cryptography, but it holes the software envelope around it. Anthropic points to vulnerabilities around widely used cryptographic components that govern how keys are generated, stored, and used in the real world – where the practical risk to financial institutions lives. Quantum risk still remains but mitigation strategies about designing quantum algorithms to safeguard data is going to be folded into the immediate challenges of AI.

Red team in a box

The good news is that Mythos is also a senior red team in a box. It can run massive enterprise-network simulations, chain exploits, escape sandbox environments, and figure out how bad guys infiltrate or exfiltrate systems. It’s pretty awesome.

But it means banks and cybersecurity experts can no longer work around episodes in the calendar. Tests, exercises and assessments were never precise but they were built against human hackers, not AI. And the resources required to attack an institution or protocol are now available to almost everyone.

And security isn’t just about the firm, the team, or the software. It’s now an industry risk, a systemic vulnerability. Shared dependencies such as open-source libraries, foundational middleware, monitoring agents, circuit breakers, and common subscription-software components are beyond the control of a given user, or vendor. Anthropic has launched a project called Glasswing to serve as a community response, and no doubt vendors and cyber specialists will introduce their own.

Every enterprise has been hacked. Boards and execs have learned to assume a determined attacker can penetrate their systems, at least to some extent. The playbook has been to shorten the distance between discovery and mounting a defense. That’s still true, but the timeframes have shrunk. We don’t yet know by how much. We’re going to find out.

Expect to hear “resilience” in every CEO and conference-stage speech. Okay, sure. What else? AI has already been challenging what constitutes value, to an enterprise or to a fintech startup. Mythos amplifies this. Having AI to detect fraud is fine, but it’s now hygiene. Enterprise value means proprietary data sets, regulatory licenses, and a governance structure that enables safe and durable human-to-machine systems.

What to ask

Surviving and thriving in this next chapter of AI will require investment into governance and systems, and not every team or institution can or will make that investment. Some may find their legacy systems too cumbersome to patch at Mythos speed. Others won’t make the cultural adjustment. Again, nothing new, but many institutions survived static mindsets because the pace of change was tolerable.

Boards, founders, and C-suite executives should ask themselves what they should expect to find should a Mythos-class model analyze the entire stack tomorrow: code, infrastructure, dependencies, vendor interfaces. How quickly can those bugs be squooshed?

They should ask where and when they depend on other companies to do the patching, like in an open-source library. Maybe the firm can, and should, organize assistance?

They should immediately review all key vendors, such as cloud providers, core banking platforms, card processors, custodians, and crypto services such as bridge providers and pricing oracles, with regard to their AI security posture and disclosures.

They should develop a PR strategy to control messaging around what vulnerabilities may exist, and find the balance between transparency and prudence. Communicating with regulators, counterparties, and customers, not to mention employees, will be important, and firms need an instruction kit around how to break the glass when there’s a fire.

And they should ensure they have governance around using their own AI tools. The nature of LLMs is they are black boxes. When a crisis hits, responses will be in machine time. What’s best practice now?

Barry Eichengreen’s new book, Money Beyond Borders: Global Currencies from Croesus to Crypto, spans from Athenian owls to the digital reniminbi to tell the story of international currencies. Emphasis on “international”: his interest is what makes a currency relevant to people and institutions across borders. It’s also a narrative toolkit for thinking about how and why a currency becomes global, and whether the dollar’s number is up.

A prominent American economist and economic historian, Eichengreen begins in Classic Greece, where Athenian owls, tetradrachm coins whose silver content and design barely changed for centuries, became universal within the ancient world because the issuer combined commercial reach, monetary stability, and credible political power. Athens had rich silver mines, a navy, and a network of subject allies; its rulers kept the coin’s quality steady, then mandated its use inside its empire, turbocharged by Alexander the Great’s vast conquests. That mix of reputation and coercion, Eichengreen suggests, is the original “exorbitant privilege.”

From there he moves through the denarius, the Byzantine solidus, and the Islamic dinar. The pattern repeats: trade routes open, a state builds administrative capacity, standardized coinage underwrites commerce, and the currency leaks across borders as a trusted unit of account and store of value. Similarly, international currencies are coalmine canaries of a great power on the wane, whether via debasement, loss of access to mines, or old-fashioned complacency.

Turning points

The hinge to the modern world comes with Florence’s florin and the Renaissance of credit. Here Eichengreen pivots from metal to book money: Florentine houses use double-entry bookkeeping and bills of exchange to turn the florin into a banking unit that often moves without any coin at all. The lessons pertain to the Dutch guilder, London’s sterling bills, and the Eurodollar market: what matters is the institutional complex around a currency: its legal protections, market infrastructure, and a deep, diverse investor base, whose confidence depends on a backstop or lender or liquidity provider of last resort.

Unlike other books about monetary history, Eichengreen is fluent in the detailed mechanics of how financial systems evolved to meet cross-border needs, first to facilitate and settle trade, and then to use the currency to develop purely financial instruments. He combines the sweep of millennia of global history with the nitty-gritty of, say, New York bankers’ acceptance notes, while keeping the book concise.

It is this grasp of financial detail that makes Money Beyond Borders a definitive account. Eichengreen has written other books on topics such as the dollar’s challenge to sterling in the 1920s and its retreat in the 1930s, but here he puts such action in a broader context, asking more fundamental questions.

He brings this insight to Nixon’s 1971 suspension of dollar-to-gold convertibility, one of the landmark decisions creating the neoliberal order that continues to shape our societies. Of note is that many smart people wrote off the dollar in the wake of the end of Bretton Woods, and Eichengreen shows how US economic heft, Treasury market depth, open capital markets, and US military power not only supported the greenback, but enhanced its primacy.

The dollar and US power were also questioned in the aftermath of the 2008 financial crisis, which led to China and others to actively seek alternatives. Yet Treasury learned to turn the dollar into a chokepoint of remarkable power.

Dollar doubts

Today, however, the excesses of that power are one reason why questions about the dollar’s primacy are more pointed than ever.

From Eichengreen’s 2,500 years of monetary history, we can see a pattern: international currencies belong to strong states with administrative capacity, fiscal resources, and a credible ability to defend their institutions, in particular the institution that bestows credibility upon the currency. That is a twofold process: technical, from Roman bureaucratic oversight of mints, to the independence of the Federal Reserve; and geopolitical, from the military expeditions of the Dutch East India Company to Britannia’s ruling the waves.

Dominant currencies have been mostly associated with republics or democracies, because they offer rules that protect property, checks and balances on tyrannical whim, and treat foreigners equally. This isn’t always true: the long dominance of the Mexican dollar had to do with Spain’s felicitous windfalls in the New World, but these ‘pieces of eight’ are also the one example of a global currency that became truly cosmopolitan, delinked from the shrinking power of the Spanish empire. A prospect that Eichengreen missed is whether stablecoins could augur something similar for the dollar.

One reason why silver dollars endured long after Spain’s imperial heyday was because of another iron law for an international currency: it needs market infrastructure and liquidity. The bill of exchange only became a truly international instrument once cities like London built acceptance houses, specialist brokers, and a central bank ready to backstop the market in a crisis. Today, that logic underpins his discussion of the dollar’s framework of New York’s money markets and Fed swap lines, versus China’s efforts to build rival settlement systems, designate renminbi clearing banks, and offer renminbi swaps to foreign central banks.

Today there is no compelling case to assume the world will dump the dollar for renminbi or euros. But foreign investors will accept only so much degradation of US management before something dramatic causes a mass switch. Financialization has eroded previous regimes, and it’s hard to imagine a society more financialized than today’s America, where savings and investment institutions have been suborned into casinos.

But these are the results of policy choices, and policies can change. It’s hard to imagine anything sensible emanating from the Trump administration, but hollowing out takes years, if not decades. The US can still remember how to manage its affairs more prudently. It’s happened before.

Pace of change

China is developing parallel markets, institutions, and its e-RMB. Eichengreen notes that the costs of using the renminbi are considerable, so that only countries such as Russia that are banned from SWIFT and the US banking network will use them to a meaningful extent. The dollar’s network effect is too strong, for now.

But just as the US might, or might not, reform its economic policies, so too could China. It’s hard to see Beijing accepting the tradeoffs necessary to make the renminbi an international currency beyond a small group of countries with few options. Eichengreen explains why the use of eRMB will still require dollars once foreigners want to repatriate their money. He also doubts CBDC network projects such as mBridge will scale.

But China is making the most of Trumpian chaos to make the case that it’s the adult in the room. It’s offering a yield on its eRMB to tempt foreign banks to hold it. The European Union could also change its stripes; although Eichengreen is skeptical, the EU’s member countries may yet take financial and fiscal integration seriously, which would make the euro a credible player.

Although Eichengreen examines stablecoins, he does so in the context of dollar dominance, not as the wedge leading to a bitcoin nirvana. Despite the book’s subtitle, he only mentions bitcoin once, curtly dismissing it and its ilk as too volatile to serve as a unit of account, a store of value, or a means of payment. Any credible monetary experiment must solve for who bears the risk, who provides the backstop, and the politics and military might behind the promise. Digital assets are relevant as reflections of central bank money (although, see financialization, above).

Should tokenization platforms become important means through which global investors hold dollars, then the Fed, the US Treasury, and the court system will have to ensure the governance, including operational reliability. Regulated and licensed stablecoins could well lead to assumptions of a Fed backstop to prevent a run. Blockchain won’t count if American commitments to rule of law, independent central banking, separation of powers, and support for foreign partners decay too far. Ditto for Chinese digital infrastructure.

What could be different this time? Speed. It took decades for Britain to pass the baton to the United States, from the shock of World War I to the humiliation of the Suez crisis. Before the second Trump administration, it was reasonable to surmise any transition from Washington to Beijing would also take decades. But now things are moving quickly. The internal constraints of the Chinese Communist Party are now the bigger obstacle to regime change.

So far we have some evidence that global investors and foreign central banks are fiddling on the margin, such as their headlong rush into gold (Eichengreen wrote an influential book, Golden Fetters, about how flaws in the gold standard led to the Great Depression). But these are indicators of concern, not of paradigm change.

Eichengreen’s sobering conclusion is that the 1930s provide a scenario in which sterling’s global role crumbled before the US was ready to take on the role of lender of last resort. Multi-polarization and similar language, were it to come about, would not mean a chummy world in which everyone gets to trade freely in their currency of choice. It means less trade, declined foreign reserves among central banks, and less growth or a worldwide depression. As painful as it is to watch the clown show in Washington, the alternative to de-dollarization is worse.

Join Jame DiBiasio on this brief exploration of the Dojima Rice Exchange.

A pair of new protocols – one for cards, one for crypto – have emerged in early 2026 to compete as the dominant standard for payment agents, in which execution transactions is outsourced to AI.

The agentic payments stack is predicted to be huge. Gartner Group estimates agentic B2B payments will hit $15 trillion by 2028; even if that’s rosy, it’s likely to match or exceed Stripe’s 2025 $1.9 trillion in processing volumes, or carve a percentage of Visa’s $14 trillion.

These emerging protocols – well, they’re not exactly new. They’re both based on longstanding browser code. The best known of these is “404 Not Found”, which is the HTTP error code that flashes when an internet user hits a dead link. The 404 is just how the servers talk to each other.

The one for payments is “402”. It dates back to the 1990s when developers assumed money would one day flow like data (the thinking at the time was about micropayments) and they reserved a status quo, 402, for “Payment Required”, in order to access an article, an image, or an API call.

That vision has only now materialized. Credit cards filled the void, but fixed transaction fees by processors negated the possibility of tiny online payments. The economics of the internet shifted to advertising and eyeballs; if information was too difficult to pay for, then “information wants to be free”, as the early-2000s slogan went.

x402 v MPP

Fast forward, we now have digital money that can move in tiny increments, AI agents that can evaluate and authorize purchases, and massive demand to move value like data. Firms have dusted off your daddy’s 402 code and are repurposing it for new money rails.

In the crypto world, Coinbase is pushing its x402 protocol, with stablecoins as the unit of account, via a blockchain rail such as Base, Ethereum, or Solana. The client provides a cryptographic payment proof, a faciliator checks the chain, and returns another code (200) to confirm the money landed.

In the cards, world, Stripe is pushing MPP, Machine Payment Protocol, which also starts with HTTP 402 code. But instead of paying each request directly onchain, the agent opens a session: a payment channel for pre-funding a balance, and then streaming micropayments off-chain via signed messages, with periodic or batched settlement. Stripe has built its own blockchain, Tempo, where users can deposit funds in escrow accounts. Although Tempo can process stablecoins, it is also designed to abstract credit and debit cards, buy-now pay-later apps, and digital wallets. An agent paying USDC or a corporate Visa card can hit the same endpoint and get the same receipt.

Neither of these means of accepting payments addresses ongoing problems with stablecoins (no recourse, bearer instruments that don’t enable money creation, cyber security questions, etc) – but assume their adoption.

Above and below

These 402s don’t exist in a vacuum. Below them sit various chains and rails: Tempo, Base, Ethereum, Solana, Visa and Mastercard networks, and real-time payment systems.

Above them run orchestration layers: Stripe’s ACP or Google’s AP2, plus newly minted “agent credential” platforms such as those introduced by Visa and Mastercard. These define how agents discover services, exchange mandates, and delegate authority, across any rail or chain.

Although these orchestration tools have their own designs (one for abstracting credit-card numbers, another for empowering e-commerce tools, etc), they are composable. They can be created and recreated to fit the need. In theory, a full agentic payments stack can cross many rails and chains to carry out a single transaction.

The best thing about the 402 trend is that it assumes open loops, not walled gardens. This has ramifications for banks, fintechs, and crypto players. Obviously those keen to operate walled gardens will find 402 a challenge. But it also means there will be places to participate and thrive.

Banks: trust anchors

For commercial banks, the advantages are clear: licenses and customer trust. In the new world of agentic finance, banks can position themselves as a credential and compliance anchor.

Banks will still be issuing credit cards via Visa and Mastercard networks, but these new cards will be virtual and rely on agent-specific tokens on the processors’ agent platforms. The banks, as issuers or acquirers, can defined agent limits, program allowable merchant categories, and set other granular controls. This gives the banks lots of scope for product innovation: a corporate card user can get a procurement agent with a budget cap, spendable on pre-approved SaaS APIs among approved vendors, in certain jurisdictions.

And it happens within their KYC’d world. For example, in Europe under MiCA regulation, agents start to resemble payment initiation service providers. Banks can support them by serving as compliant x402 facilitators, stablecoin custodians, and embed MPP sessions (escrow accounts) into their existing services for businesses small and large.

Few banks are ready. They need to invest in agent identity and reputation systems, and systems that can communicate with orchestration layers. They also need a regulated crypto and stablecoin stack to handle custody, on/off ramps, and reporting.

This is surely the direction of travel – if the regulators are along for the journey. Agents allowing a human to click “pay now” will only go mainstream if questions about liability and consumer protections are hammered out. Regulators will be keen to ensure their commercial banks can become trusted wrappers around permissionless protocols or, in Asia, atop domestic fast-payment rails such as India’s UPI or Singapore’s PayNow. This would give leading banks the chance to become national or regional hubs for agent flows.

Fintech agility

Banks’ opportunity is clear, partly because of their licensing. Fintechs are more agile, and can position themselves throughout the agentic payments stack.

They could be payment processors, packaging x402 and MPP for merchants and developers who don’t want to think about sessions, chains, and the rest. It’s all about abstracting cards and wallets, and connecting to every protocol, to maintain a clean, simple user interface for merchants.

Or, fintechs could climb up a rung and become super orchestrators, competing against Stripe and Google and the card networks. There’s a business model for helping customers choose the optimal payment method based on cost, latency, risk, user preferences, and regulatory constraints. This has been a viable model for fintechs in the pre-agentic world, and there’s no reason why some can’t make the leap.

Similarly, there exist fintechs who own niches, and this model can extend to agentic payments. That might mean compliance and identity, risk and fraud prevention, or “compliance as a service”.

Merchant visibility

For merchants, the urgent task is to be visible to agents. If a consumer’s search is embedded in a payments agent, then merchants need to have bank and fintech providers, or their own capabilities, to win the attention of the bots. For starters this could mean being plugged into platforms such as Stripe ACP or Google AP2, or card networks. The big service providers will be eager to make this process easy so that merchant catalogs and prices are available as structured data, not just HTML.

Eventually, merchants will want to build their own retail agents. This is the new branding, except it’s via agents rather than ads. Old-school marketing still has a role to play, to ensure the business is kept on programmable approved lists. But merchants will explore new business models around products that attract machine attention, and could lead to the sort of micro services (and micro payments) that internet HTML developers first envisaged in the 1990s.

Banks, fintechs, and merchants still need a few common jigsaws to complete the puzzle. Agent intention that is universal will require standards. The regulations for wrapping open rails such as x402 need to be agreed. The settlement infrastructure for high-volume micropayments has yet to be refined. And most of all, credentials and histories need to be made portable and reusable, be it cryptographic tools such as zero-knowledge proofs, or more traditional identity verification that can be fitted to agentic payments.

That’s a lot of work, but the prize is the trillions of dollars expected to flow through the agentic payments stack within a few years.

The global stack

Nor is this an opportunity exclusive to the United States – notwithstanding the emergence of x402 and MPP by Coinbase and Stripe in the US, along with LLMs such as ChatGPT, licensed stablecoins such as Circle’s USDC, and US regulatory debates.

The trend of needing a way to pay other machines at high frequency, often in tiny amounts, under programmable constraints, is global. Indeed, early experiments on AI agents paying for real-world goods have taken place between Santander and Mastercard in Europe.

Nonetheless, geographic differences are a reality. Banks and fintechs in Asia (east and west) and Europe are already well versed in tokenizing credentials, running real-time payment rails, and navigating complex cross-border regulation – a real weakness for US firms. Regional orchestration and compliance will provide business opportunities. Domestic payment systems such as UPI , Brazil’s Pix, or KakaoBank in South Korea are adept at high-frequency, low-value payments, but they need to add agent-friendly interfaces, 402-style semantics, and standardized mandates. A fintech that provides a “UPI for agents” can become the local hub, while routing bigger transactions over stablecoins or x402.

Local and regional crypto players have a huge opportunity to break into new merchant relationships where small, cross-border payments are critical, and where banks are scarce. Singapore, Hong Kong, and the UAE have arguable friendlier (or at least clearer) stablecoin regimes than the US. These financial centers should become hubs for agent-native wallets, stablecoin settlements behind card programs, and on-chain identity services. But whether it’s in Europe or these small financial centers, permissionless protocols will still need permissioned wrappers. Someone on the ground needs to hold that license.

We have more wallets, QR schemes, blockchains and tokenized assets than ever before. Yet the world’s most reliable ways to move value across borders are still decades‑old card schemes and correspondent banking networks. The problem is not a lack of innovation; it is that the innovation is splintering into incompatible islands.

Until that changes, new rails will keep pressuring traditional finance at the margins without fully displacing it at the core, leaving legacy players free to plug’n’play with fintechs without eroding their margins. There’s nothing wrong with legacy payment processors and banks continuing to prosper, but fintech’s promise of competition is being diluted by fragmentation.

On public blockchains, the basic economics of decentralized consensus actively push toward fragmentation. The Bank of International Settlements recently noted that validators charge higher “coordination rents” to chains that are more decentralized and secure, resulting in congestion and gas fees. Spikes in fees on Ethereum – the fave of financial institutions – leads other users to migrate to less-trusted venues such as Solana or Tron. This has competitive merits within the context of blockchains, but it splinters activity and liquidity across dozens of settlement layers and layer-2 chains. This inhibits the ability of stablecoins to serve as fungible tokens without expensive and risky bridges.

Neobanks are growing fast, with the majority of consumers globally now maintaining relationships with two or more providers, according to Standard Chartered. But this isn’t a case of new global banks that compete with the likes of StanChart. At a regional level, this proliferation is fragmenting account balances, creates redundant KYC, and leaves users with a haphazard set of experiences and apps.

There are terrific examples of regionalism: the rise of mobile money wallets in Africa, and efforts in Southeast Asia such as Project Nexus to knit together QR-code based accounts. A recent report by the International Monetary Fund looks at Thailand and finds QR payments are beginning to substitute for traditional banking and card payments, which is bringing financial access to many SMEs.

But African wallets don’t integrate with card schemes and traditional bank accounts at the regional level – a problem for many SMEs. In Asia, QR operability for wallets is entangled in bilateral, national standards, with local regulation and currency regimes throwing sand in the gears. The IMF says Southeast Asia needs to beef up oversight and reporting to enhance trust and security of digital payments across QR-based systems.

Corporates and treasurers face similar issues. Real-time schemes, virtual accounts, digital wallets, tokenized deposits, and embedded rails are all great. They are designed to expand liquidity. But they are mostly built in silos, across varying cut-off times, costs, FX exposures, and regulations. More liquidity, maybe – but sheared off in different places.

TradFi’s endurance

Add it up, and this explains why TradFi is able to retain its dominance in cross-border finance: global acceptance, fungibility of money, dispute rights, and fraud and compliance frameworks trump new tech and digital-first business models. That’s true even when the new systems are cheaper: in fact, lower costs for new rails don’t reflect the way TradFi bundles complexity within its services. That complexity doesn’t disappear by adopting blockchain, wallets, or neobanking.

Would-be fintech disrupters have developed three playbooks to address this problem.

First is orchestration. This has been around for a long time. Many fintechs have emerged over the past decade to help treasurers and SMEs juggle multiple rails: Adyen, Stripe, GrabPay, and Payoneer for merchants, Wise, Revolut, Airwallex and Nium for SME and retail. PPRO, Rapyd, Remitly, TerraPay, and Thunes orchestrate local methods for emerging markets. In the US, Plaid provides interoperable layers for data and account-to-account payments, as TrueLayer does in Europe.

In crypto, Circle orchestrates across chains and into banks for stablecoins. Canton Network synchronizes tokenized assets and workflows across permissioned DLTs favored by financial instituitons; Partior does this for fiat FX settlement in tokenized form.

But large banks have a natural advantage. Many banks have adopted ISO 20022 standards around transaction messaging to make correspondence networks capable of handling digital rails. They have huge teams doing API connectivity. And they know how to bundle and charge for multiple payment methods.

Orchestration at scale requires the kind of internal investment, integration with other client systems, and data standards and digital identities. Banks aren’t necessarily better at these things, but they know how to proceed in a compliant manner (which is why they prefer to deal onchain just among themselves, hence the growing usage of Canton). The object is to mask the complexity for the most valuable corporate clients.

A second approach is top-down, relying upon an institutional anchor such as central banks. Their concern is the singleness of money, which has relied on central banks to backstop market and payment systems. This creates a shared anchor. Without it, private monies and networks tend to fragment, as stablecoins do today.

The emerging solutions include multi-CBDC platforms, such as mBridge, or other regulated frameworks for stablecoins and tokenized deposits. While mBridge is live, it remains limited mainly to RMB-related transactions, despite its multilateral backing. There remain questions about interoperability at scale, governance, and control.

The third play is to develop federated wallet networks. The leading example is Ant Group’s Alipay+. In mainland China, Alipay has built an integrated ecosystem of payments, savings, credit, investments, and insurance around a vast user base. Abroad, Ant has partnered its way into compatible infrastructure.

Through equity stakes, its tech powers local wallets, such as GCash in the Philippines or KakaoPay in Korea. Alipay+ allows these wallets to be mutually interoperable at point of sale across more than 100 markets. Ant handles routing, FX, and settlement for millions of SMEs, building a global network that avoids card schemes altogether. Alipay+ follows its users within their existing financial apps rather than trying to get them to onboard a new one. While Alipay+’s coverage is not as global as a leading bank’s, or a Visa/Mastercard’s, it looks like an emerging competitor at scale for retail and small B2B flows.

Ant Group has never focused on large-scale capital flows or corporate treasury, however, so it’s unlikely this approach would work at that level – we’re back with our hodgepodge of fintechs, digital assets, banks, and cards.

Singleness of experience

To sum up, the best solutions to fragmentation are building networks around existing user flows without trying to rewrite local regulations; bank-driven multi-rail orchestration on the back of APIs, ISO 20022, and real-time connectivity; and institutional trust anchors for digital settlement, notably in permissioned DLTs.

There are more solutions being tested. Might tokenized assets become embedded in mainstream treasury operations? How much potential remains for corporates to exploit the data structures of ISO 20022 messaging? Can national identity and privacy conditions be extended into cross-border flows?

What has long been understood is that market forces alone don’t lead to convergence to a single winning rail. That was also true with correspondent banking. Market forces and regulatory agendas continue to drive the proliferation of chains, wallets, and domestic schemes, with harmonization possible but complete integration unlikely.

Fragmentation is the feature, not the bug, in cross-border finance. Fintechs and blockchains have emerged as new ways to manage or hide this complexity, via federation, orchestration, or atop regulatory anchors. Systems that preserve the singleness of money – which is to say, a single user experience across underlying rails – are more likely to succeed than rickety reliance on bridges and other new versions of middleware “spaghetti bowls”.

The singleness of money is often presented as an obsession of central bankers, which can be ignored by tech startups, but this is a mistake. It’s a competitive necessity. As the world seeks alternatives to US dollar transactions, the costs of fragmentation are going to rise. Who’s going to solve this problem?

Existing commercial banks and payment schemes continue to serve as backbones, and will not disappear, because they remain best placed to do so. Despite the threat from platforms – fintechs, wallets, and crypto providers – to consume the value-add, legacy businesses that solve for the intensity of fragmentation can avoid becoming dumb pipes.

Where fintechs have excelled is in a singleness of experience: taking a consumer or SME corridor and integrating travel, remittances, and e-commerce with credit, rewards, and investments; by embedding financial services within broader activities.

This is the challenge for tokenization and digital assets. Within DeFi native experience, the models are proven, but connecting these to the real world of business and income streams requires an ability to manage fragmentation that has yet to be demonstrated.

It’s a Monday morning in the spring of 2030 and Marco, the treasurer of a large, AI-heavy multinational enterprise, starts his week by opening a liquidity dashboard that looks quite different to the one he used a decade ago.

Cash now comes in four categories. Marco surveys tokenized bank deposits in multiple currencies for payroll and bills. He checks his tokenized money-market fund shares, and his regulated dollar stablecoins for cross-border payments.

Then there’s the fourth category: compute tokens.

These are not cryptocurrencies in the usual sense. They are claims on a defined amount of computation at specific AI labs, such as Anthropic.

Marco holds them in two main forms. First, on‑demand compute balances: instantly usable claims that AI agents across the firm draw down every time they call a model. Second, term compute notes: futures contracts locking in a stream of compute over the next three years at an agreed price and service level.

Corporate treasurer’s day in the life

Every Monday, payroll drains tokenized deposits. Marco tops them up by redeeming some money-market tokens. He knows that next day, the corporation’s new AI-driven product line will require more compute, so Marco buys a tranche of term compute notes, paying with redeemed MMF shares.

Wednesday comes and a market wobble forces the multinational to raise intraday liquidity. Marco pledges some of its compute notes as collateral in a short-term repo with Alicia, a global dealer at a global bank (it could be a JP Morgan or Goldman Sachs, or 2030’s neo-bank darling). She sends him tokenized cash in return.

The next day, another AI provider raises its lab-access prices again, so Marco enters into a swap with Alicia to hedge his future compute costs. Marco ends his week by rebalancing, keeping enough on-demand compute to run the business, enough term notes to lock in strategic capacity, and the rest in conventional cash and equivalents (mostly RMB, HKD, SGD, JPY, AED, plus some dollars because the multinational has a huge US footprint, and EUR for croissants).

Compute tokens have become a working form of money for one crucial slice of the firm’s activities: everything the company does through AI. They are liquid, priced in familiar ways, accepted as collateral, and woven into the same digital rails that carry bank money and securities.

Compute as money

This idea recently surfaced by some tech insiders who think bitcoin will be replaced by “Anthropic Coin”. (Anthropic is the lab of choice because it is a pure play, independent of broader internet businesses such as Microsoft or Google.)

The argument goes like this: Large AI labs control scarce, capital-intensive infrastructure (chips, GPUs, model weights, data centers); they meter access to this infrastructure in tokens; they sell long-term capacity commitments to enterprises as well as to other machines or agents.

These tokens today have money-ish qualities: they serve as a unit of account within AI-heavy firms, with product teams budgeting in terms of tokens instead of hours or dollars; they can serve as a medium of exchange between agents; and this makes them a store of value for bots that need to plan for the enterprises’ future computing needs. Compute tokens don’t buy goods and services but they do buy cognition.

Are compute-token enthusiasts unveiling the future of finance? Not so fast. These tokens today operate more like prepaid utilities or air miles. They lack standardized legal form, robust backing structures, and integration with the broader financial system.

Some pretty big “ifs”

There are reasons why “AI money” is, for now, a fantasy. First, the argument for compute money is being framed as an alternative to bitcoin. This rather puts the cart before the horse, as bitcoin isn’t money. Scarcity alone doesn’t make it so. The flaws in thinking bitcoin is something we might all use to pay our debts apply to the argument about compute.

However, the idea of compute as money is more compelling than the bitcoin story. It’s rooted in actual supply and demand linked to the real world. Bitcoin may not be money but its invention has given rise to a new, alternative infrastructure for capital markets, one that is beginning to integrate with traditional networks, and these rails would indeed be a good fit for compute tokens.

Second, we are assuming current large-language models prove to be sufficiently reliable to anchor anything money-like. However, LLMs do bullshit, a feature that is baked into their models. Hyper-scaling compute hasn’t corresponded to similar gains in LLM robustness (perhaps it could, but the commercial models prefer BS to LLMs failing to provide a response). Trustworthy AI will need new architectures that will require time and a lot of investment to build. It’s therefore hard to put faith in compute tokens as claims on a service, let alone as money.

Third, we are also assuming the issuers of these tokens – the big AI labs – will survive. While there is real demand for AI, many serious analysts question whether these revenues will ever come close to the complex debt structures that Silicon Valley and Wall Street have concocted to keep the wheels spinning.

A huge financial crash would not end the AI story, but it could end OpenAI or Anthropic. In this scenario, compute contracts or deposits start to resemble tulips, at least until the industry recovers on a more stable footing.

Narrow banking redux

These “ifs” all rest on our final, and biggest, assumption: that compute tokens rapidly evolve into a new form of narrow banking that operates alongside TradFi.

Narrow banks are basically like today’s stablecoin issuers: deposit-taking institutions that only invest in safe, liquid assets, and do not engage in risky lending and maturity transformation.

A “Compute Bank” would accept fiat or tokenized cash from users, issue compute deposits and compute notes, and hold its assets in contracted capacity with AI labs or data centers plus fiat cash and short-term government debt, with some risk management rules resembling capital reserves, but in compute terms. The bank would earn money on transaction fees and float.

Just as we are now getting comfortable with licensed stablecoin issuers with fully backed reserves (although perhaps we are too comfortable), compute banks would have to prove their tokens are stable, predictable, and instantly redeemable for AI work.

AI labs would use compute banks as wholesale customers, selling them capacity up front, locking in that revenue, and relying on the bank to distribute capacity to the market. For traditional banks and dealers, compute deposits and notes would appear as a new class of high-grade, infrastructure-linked paper, potentially eligible for collateral in repo and derivatives.

This would not replace traditional money markets: everyone will still pay taxes, wages, and non-AI contracts in their fiat currency. But over time, if AI lives up to its promise across industries, corporate treasurers like Marco would come to hold compute deposits alongside cash and treasury bills, and dealers like Alicia would make markets in compute notes and compute-backed securities. MSCI would be rolling out compute indexes, and enterprises might begin trading in compute terms, creating a currency akin to petrodollars. All backed by productive capacity rather than faith in a government or hand-wavy belief bitcoin.

Not ready for prime time…yet

For this to happen, though, these assets have to be safe, cheap, and predictable – as T-bills or repo markets are today. Compute is none of those things. It is expensive, with mind-bogglingly up-front capex on hardware and data centers; it’s a voracious consumer of energy; and we have no model for cost per unit of useful work – how do we measure someone prompting their way to protein folds to cure cancer, versus someone burning carbon to ask their LLM about Hollywood gossip or sports trivia?

Moreover, compute carries its own operational risks: outages, cyberattacks, misuse, sudden regulatory changes, Iranian drones blowing up Microsoft data centers in the Middle East, chip makers in South Korea losing access to cheap electricity because of geopolitical hazards.

Finally, when we bring Wall Street practices into the equation, we end up importing Wall Street culture. That implies new layers of opacity, complexity, and leverage around speculative expectations of future AI demand. Prediction markets would be all over this asset class, bringing chaos as much as they might foster liquidity. The more the world comes to depend on AI, the more vulnerable it would be to a financialized compute market always vulnerable to a meltdown.

Therefore, the notion that compute deposits are comparable to Circle’s licensed stablecoin USDC has a long way to go. And even if a compute bank could claim full and reliable backing, users are still exposed to “money” that could go poof! because an LLM screws up, is lethally prompt-injected, or manipulates the human for its own arcane agenda.

What has to happen

This doesn’t mean compute money is a non-starter. It simply means a lot of things need to happen before it can take off. The incentives are real.

On the technology side, this implies the development of LLMs that don’t hallucinate or lie on important tasks – a goal that new AI labs in Silicon Valley are beginning to address. It implies strong evidence that AI systems can augment or replace human labor across many industries and activities. And it implies that humans have better control over AI, which allows regulators to regard it as imperfect but manageable.

On the economic side, we’d need to see AI leading to sustainable productivity gains across many industries, so that AI labs enjoy healthy margins. We need these companies to be bedrocks of stability, not debt-fueled tinderboxes.

On the institutional side, we’d need transparent, simple, and auditable structures for compute claims. This goes hand in hand with clear legal terms and sensible backing. The stablecoin example is probably insufficient, as it remains riddled with weak attestations: this isn’t good enough if compute tokens are to become the new money. We’d also need regulatory regimes that know how to supervise compute banks and enforce capital, liquidity, and risk-management standards. The upside for the compute banks would be the opportunity to integrate with traditional deposits and commercial-bank money.

These developments are all possible, although they require multiple changes of direction in Silicon Valley, on Wall Street, and in the halls of regulators and policymakers. But it’s an optimistic vision, for the emergence of compute money implies a successful adoption of AI across industries and society. The horror story is also possible, of compute money gone haywire, but it’s more likely that AI will self-destruct (either in economic and valuation terms, or something worse) before computer money has a chance to take hold.

Indeed, the idea that AI companies can morph into bedrock financial institutions, but only under the right conditions, might be the incentive we need to get them on a more sustainable, less dangerous track.

Finally, what about VC firms themselves? What’s it mean for a VC to become an AI-first financial firm? And going back to the blockchain theme, does tokenizing a VC’s own funds generate value?

Dušan Stojanović is Singapore-based founding partner at True Global Ventures, a global firm focused on seed and Series A opportunities. He describes himself as a “global citizen with a Swedish passport”. He’s been investing in AI since 2012, and in blockchain since 2016. He helped found TGV in 2010 and is now on his sixth fund.

Timecodes:

0:00 - Dušan Stojanović, True Global Ventures

02:09 - Stablecoin/Web3 convergence with AI

6:15 - OpenClaw’s “ah-ha” moment for SaaS companies and its impact on TGV’s portfolio of companies

9:46: What is an “AI-first” startup? How can you tell when “AI” is a real differentiator or a marketing ploy? What’s “AI first” in a regulated field like payments and finance?

12:25 - How do companies risk-manage the AI itself? And how Dušan looks at valuations in AI companies.

15:27 - What’s a defensive business moat for an AI startup today?

17:35 - TGV’s investment strategy: stage, size, how to compete against the biggest Silicon Valley names as the landscape for venture changes

22:46 - How “AI first” applies to TGV’s own business

25:11 - The geopolitics of investment for TGV

26:50 - Exits and liquidity

28:01 - Will TGV tokenize its own funds?

In Part 1 of this series, we looked past “trillions settled on‑chain” and asked where stablecoins actually matter in payments. Part 2 walked through how lawmakers are forcing these instruments into familiar legal boxes, from the US Genius Act to MiCA and Asia’s first licensing regimes.

Part 1: Stablecoins and the real world of payments

Part 2: The legal limbo of stablecoins

This final piece tackles the awkward question: if we now have 1:1 reserves, licenses, and real users, what can still break the peg?

The answer sits in a narrow strip of infrastructure that connects stablecoin treasuries to the rest of finance: repo markets, Treasury liquidity, and the capital that dealers and issuers are willing to put at risk. That strip – perhaps “knife’s edge” is a better description – is where stablecoins live or die in a crisis.​

Solvency vs liquidity

New rules are explicit about solvency.

The Genius Act forces US “payment stablecoin” issuers into a tight reserve box: cash, bank deposits, and short‑term Treasury‑linked instruments, at least one‑for‑one against tokens, with attestations and audits for larger names. Hong Kong and Singapore take a similar line, insisting reserves sit in segregated bank accounts and low‑risk securities.

A quick explanation of repo may help.

For large banks and dealers, repo is the main way to turn securities into cash day‑to‑day. It lets them finance big inventories of Treasuries and other bonds without locking up scarce long‑term funding or equity. When repo is cheap and plentiful, banks and dealers can carry more inventory, quote tight prices, and stand between buyers and sellers. This makes overall market liquidity look deep and smooth. When repo funding tightens, they have to shrink positions and widen spreads, and even “safe” assets can suddenly be hard to move without hitting the price.

A repurchase agreement, or repo, is structurally a secured loan. One party “sells” a security (usually a government bond) to another, while simultaneously agreeing to buy it back later at a slightly higher price. Economically, the first party is borrowing cash and posting the bond as collateral; the second is lending cash against that collateral. The price difference is the interest on the loan, and the haircut – the amount by which the collateral value exceeds the cash lent – is the lender’s buffer against market moves.

Looked at from the other side, a reverse repo is what the cash‑rich institution enters into: it is the lender, taking collateral and earning a short‑term, money‑market yield. For that lender, whether a bank, money‑market fund, or stablecoin issuer, reverse repo is a way to park liquidity very safely and very short‑term, while holding high‑quality collateral that can be sold if needed.

Taken together, repo and reverse repo form the core of the secured funding market. When those markets are healthy, banks can fund their securities books smoothly; when they are stressed, the funding chain that underpins both bank balance sheets and bond‑market liquidity comes under strain. The most dramatic example of this was in 2008, when the rest of Wall Street made a disastrous bank run against Lehman Brothers via the repo markets.

So, back to our stablecoin story. Legislation and regulation in leading money centers have addressed the obvious failure mode: the issuer that mis‑invests or misappropriates assets. But these rules don’t say what happens if those reserves can’t be immediately turned into cash, which is likely to happen when everyone heads for the doors at the same time.

Three design facts define this liquidity risk:

• Hierarchy. Stablecoins sit below banks and central banks. Redemptions settle in commercial‑bank deposits, not reserves, and often require selling securities through dealers. The peg is hostage to market access.

• Portfolio shape. The big issuers are conservative: treasuries full of T‑bills, overnight reverse repo, and bank deposits, designed to minimize duration risk. That shrinks insolvency risk but pins them to the liquidity of Treasury and repo markets.​

• Intermediated redemption. Even in a Genius world, direct token mint/burn tends to be limited to a narrow circle of institutions, with everyone else relying on secondary markets. MiCA’s direct‑access requirement simply shifts where pressure shows up: from secondary‑market discounts onto stablecoin issuer balance sheets.

In fair weather, these choices look prudent. Under stress, they turn “fully reserved” stablecoins into forced sellers in markets that may not want to buy.

The redemption wave

In the process of token mint‑and‑burn accounting, deposits move between accounts, but the system does not conjure or destroy bank money just because someone flips between tokens and deposits.​ Stablecoin issuers are, in this sense, “narrow” banks that don’t create money.

That remains true in a run. What changes is who is forced to hold which assets, at what price.

For a detailed look at repo mechanics, check out Austin Campbell’s articles at Zero Knowledge Consulting. He’s a good resource on this topic. But simply, a stylized wholesale redemption works like this.

1. A token holder sends stablecoins to the issuer to redeem.

2. The issuer sells T‑bills or unwinds reverse repo via a dealer to raise deposits.

3. The dealer’s bank moves reserves to the issuer’s bank; the issuer’s bank credits the issuer.

4. The issuer wires deposits to the redeemer’s bank; the tokens are burned.

On paper: fewer tokens, the same aggregate deposits, a reshuffled distribution of Treasuries and repo positions.

In practice, each hop can misfire:

• Dealers can hit internal risk limits or leverage constraints and widen spreads rather than warehouse more paper.

• Repo haircuts can jump, squeezing some borrowers out just as issuers most need liquidity.​

• Trading and settlement systems can slow or fail under load, delaying execution.

The March 2020 dash‑for‑cash, when Western countries were shocked by Covid and everyone on Wall Street grabbed cash at the same time, showed how little selling it takes to flip into this regime. A few hundred billion dollars of Treasury liquidations in a multi‑trillion market were enough to dislocate prices and force central‑bank intervention. Stablecoins were marginal then. With hundreds of billions in reserves and growing real‑economy usage, they are large enough now to be part of the next dash‑for‑cash.

Given current wobbles in capital markets – AI bubbles, Iran war/oil dislocations, private-equity troubles – such a crisis is very possible. Worse, we have the example of the run on Silicon Valley Bank, which transpired too quickly for the Fed to backstop SVB. Now imagine that kind of one-click panic involving a stablecoin issuer holding billions of dollars worth of US government debt.

Safe harbor or choke point?

Reverse repo has become the focus of reserve disclosures, and not by accident.

A recent paper by MIT Media Lab, “The Hidden Plumbing of Stablecoins: Financial and Technical Risks in the Genius Act Era”, shows that an overnight reverse repo with a haircut protects an issuer against rate moves better than owning longer‑dated bonds outright. The lender is over‑collateralized; even if prices fall between the two legs, they can sell the collateral and recover principal up to the haircut. Short tenors or frequent re‑margining let them roll into higher yields rather than sit on underwater positions.

With stablecoins, the issuer takes a bank deposit, lends it overnight via reverse repo to a dealer, receives Treasuries as collateral, and either gets the cash back or sells the collateral if something breaks. Deposits move between balance sheets; nothing is created or destroyed.​

This sounds safe, but in a crisis, the protection reverse repo offers on individual balance sheets can become a bottleneck at system level.

• Cash lenders may pull back or limit business to the strongest counterparties.

• Haircuts can widen, forcing more collateral for the same cash.

• Central‑bank standing repo facilities backstop only a small club of firms, under their own policy and risk limits.

Stablecoin issuers are not in that club. They access repo through the same big banks and dealers that, in a crunch, can hit various regulated constraints on how much capital banks must hold to operate in “low-risk, high volume” activities such as repo that were imposed after the 2008 Lehman bankruptcy. Genius instructs supervisors to write capital and liquidity rules for issuers, but those standards are still in development; for now, issuers run with thin capital relative to their potential obligations.​​

Reverse repo is therefore a private buffer that works well in normal times. It is not a magic shield against a broad funding crunch.

Who eats the loss?

As noted above, the US, Europe, Hong Kong, Singapore and other places are introducing capital rules for banks and brokers meant to encourage the safe handling of stablecoins. Better capital doesn’t stop runs, but it changes who can absorb what.

A well‑capitalized issuer can redeem for longer out of bank deposits, accept more slippage on asset sales, and survive operational losses without dipping into reserves. Basel capital regulations have until recently forced massive over-collateralization against stablecoins. Standard-setters like the Financial Stability Board (created by the G20 after the 2008 crisis) insist intermediaries serving stablecoins should hold capital for operational and liquidity risk, not just for the underlying credit exposure, which penalizes token-related business.

These buffers are being whittled down by crypto-industry lobbying. The US Securities and Exchange Commission has now decided broker-dealers can treat stablecoin capital requirements as equivalent to cash, with a mere 2 percent capital reserve requirement. This will encourage them to hold stablecoins and use them for settling client trades or for internal liquidity “sweeps” (automated movements of extra cash into yield-bearing investments). This makes stablecoins far more attractive tools for corporate treasurers and payments.

But it doesn’t change the basic reality that when a dealer needs US dollars to meet a wave of redemptions, it still has to turn T-bills or reverse repo into bank deposits via the same constrained balance sheets and repo markets. There may be some benefit to encouraging broker-dealers to inventory stablecoins, which can provide a broader market, but whether this makes a difference in a crisis is hard to say.

If there’s a dash for cash, no one is going to be asking “what reserves back this token”? They’ll be asking “whose balance sheet stands between me and the exit”? Which banks will be able to ride out a crisis, and which will be forced to pull back from markets just when treasurers need them the most?

Better, safer

Ultimately, the risk has to sit somewhere. That could be on the balance sheet of thinly capitalized issuers, who become forced sellers in a crisis. It could be on the balance sheet of large banks and dealers, whose own capital and liquidity constraints could turn them from serving as shock absorbers into crisis amplifiers. Or it could even sit on central-bank balance sheets, via banks’ capital reserve accounts or (in theory) direct token settlement.

If we think a 1:1 reserve rule solves the hard questions of liquidity, then we are deluding ourselves. This does not mean stablecoins are “bad” or that banks shouldn’t be willing to deal in them. Stablecoins offer a rare opportunity to bring real competition to the giant Wall Street banks that exercise far too much power in US politics, and therefore in the world. But if crypto tools are going to become a systemic part of financial fabric, we must ask far more of issuers and exchanges, as well as traditional institutions.

What might this look like? Well, here are three suggestions guaranteed to be widely hated. But here we go:

1. Boost capital requirements in repo. Focusing just on stablecoin-specific risks misses the bigger picture. Regulators can adjust capital-related surcharges so that balance‑sheet used for market‑making in Treasuries and central‑bank‑eligible collateral is not penalised as heavily as genuinely risky assets, provided banks hold strong Tier 1 capital and meet liquidity standards. For their part, banks can strengthen repo infrastructure, by dedicating capital and liquidity limits to secured‑funding desks, adding contingency plans for expanding repo and bond‑market intermediation in stress, and backing balance sheets by pre‑arranged access to central‑bank facilities where available.

2. Require stablecoin issuers build liquidity buffers. Issuers should hold a meaningful slice of reserves in instantly available deposits at strong banks and in central‑bank‑eligible overnight reverse repo, above and beyond the legal 1:1 minimum. They can also publish details about their available deposits, overnight reverse repo, and holdings of T-bills by tenor, so counteparties knows what to expect if they want to redeem. And these attestations are weak and backward looking: issuers must undergo regular, independent stress tests that replicate conditions akin to the Covid shock of March 2020 or the SVB run.

3. Regulators modernize the infrastructure. Backstops such as the Fed’s facilities for repo markets and money-market funds should also be tested for stablecoin-related crises, to ensure collateral can be turned into cash at a reasonable haircut, and build resilience into broker-dealer balance sheets. In addition to changing data and capital rules mentioned above, regulators can also consider how to improve central clearing of Treasury and repo markets, so these don’t gum up in a panic.

Yeah, thought you’d hate that. But here’s the truth: stablecoin pegs will hold, or break, not on the quality of assets in a PDF attestation, but on whether someone with enough capital and balance‑sheet room is willing to be buyer of last resort when stablecoin holders all decide they would rather have deposits today than T‑bills tomorrow. This story isn’t about the peg. What counts is the plumbing.

Part 1: Stablecoins and the real world of payments

Part 2: The legal limbo of stablecoins

Enter zero-knowledge proofs, a cryptographic technique that has a lot of promise for digital finance. It first emerged in the world of crypto and Web3, and remains central to tokenization and DeFi. But its potential application extends far beyond anything blockchain. It could in fact prove to be a crucial tool to help enable agentic AI, safely. But it could also become a new risk vector, and ZK Proofs face plenty of obstacles to adoption.

Alex Scheer is founder and CEO of zkMe, a startup that uses ZKPs to help businesses and users verify identify and asset ownership, and promotes ZKPs as the universal identity protocol for onchain compliance, risk management, and underwriting. Raised and educated in Germany, he now lives in Hong Kong.

Timecodes

0:00 - Alex Scheer and zkMe

2:30 - Privacy versus transparency, identity, and how these can help guard against deepfakes and other security risks in digital finance

4:46 - The relevance of Web3/crypto technologies in the broader world of finance and AI

6:01 - What are zero-knowledge proofs? Alex explains this tech in terms of trust, and real-world examples of ZKP in the business world

10:29 - How do we know we can trust AI agents? Alex’s different take on KYA - “know your agent” – and the lack of protection for identity in the marketplace

17:09 - ZKP adoption challenges; how ZKPs integrate (or not) with existing systems and vendors in compliance and risk management

19:36 - What’s the likelihood that ZK proofs or ZKP providers become a new node of vulnerability?

21:11 - ZKP’s potential in fostering open-finance models, and the concept of data “reusability” and making it self-sovereign

26:40 - But '“self-sovereignty” has never taken off…is this time different?

28:57 - What has to happen to get banks, regulators, and fintechs want to support the use of ZKPs? Alex’s top business opportunities.

Related conversations

Alex spoke at length about identity, privacy, and the intersection with agentic AI. You should also check out these related videos:

The AI threat to VCs, with Ray Wyand

Agentic open banking, with Jessica Liu

Cross-border digital banking, with Urzula McCormack

For stablecoins to become mainstream payments instruments requires they be compliant, but that’s just table stakes. The competitive edge for issuers is to build them with the right balance of identity and privacy standards.

Compliance is one of several aspects to building a sustainable, far-reaching financial infrastructure using stablecoins. This series of articles, “The plumbing matters more than the peg”, examines the components to realizing this goal, including reserves management, liquidity, and resilience; Part 1 already looked at the scope of opportunities for stablecoins in payments.

Part 1: Stablecoins and the real world of payments

Legal status

The good news is that many jurisdictions have legislation and regulations regarding stablecoins. This is a big change compared to about eighteen months ago, when there was little or none. However, whether it’s the US or Hong Kong, Europe or Singapore, London or Dubai, stablecoins are not legally operating at the same level as cash.

While both sides to a transaction can increasingly treat certain regulated stablecoins as money for booking and settlement, stablecoins do not have general legal-tender status.

Accounting standards view stablecoins as an intangible property or a financial asset, which requires a capital charge, whereas cash would not. More jurisdictions are interpreting their accounting standards to accept stablecoins as cash equivalents, which eliminates one barrier to their adoption. Even so, stablecoins still present merchants with an accounting nightmare: they have to track fiat fair-value receipts, and recognize any gains or losses on fiat conversion, creating additional paperwork than if they just held cash or fiat e-money.

Tax authorities mostly treat them as property or financial assets, not as cash.

Even if a regulator defines a stablecoin as a payments token, not a security, the taxman may still consider the transaction to be fair game. Spending via a bank deposit doesn’t trigger a capital-gains tax, but paying with a stablecoin still can. And tax authorities are not accepting tax payments in stablecoins, an important indicator of whether the state regards them as ‘money’.

Finally, we don’t know yet how courts and law enforcement will treat redemption at par from a clearly liable but troubled issuer; precedents around money-market funds have not been tested in a stablecoin case.

For a licensed entity to book, pay tax, and legally enforce stablecoins as a normal payment obligation depends on accounting, tax, and legal changes. Some of these shifts are in progress, others have a long way to go.

This restricts stablecoins to a narrow set of applications. As a counter-example, look at China’s eRMB. Because it’s recognized by the state as legal tender, the eRMB can be used for paying taxes and receiving government services. Stablecoins can’t worm their way into everyday life without that recognition by the state. Corporate treasurers, projected by the fintech industry to be the biggest adopters of stablecoins, confine their use to niche flows related to on-chain liquidity and collateral. They aren’t being used for general payments, let alone for mass-market invoicing or payroll.

One last challenge to settling the legal status of stablecoins is their global nature.

Global 24/7?

Even when regulators agree on sensible rules, they still only operate within their own jurisdiction. Correspondent banking evolved to meet this reality; hence the duplicative compliance checks. Crypto rails have been built to operate globally, 24/7, with little regard for regulation. Backfitting this infrastructure is easier if there is either a very big market (the US) or a bloc that offers a single license, such as in Europe’s MiCA regime.

But in Asia or Latin America? Regulation is fragmented across two dozen jurisdictions. Licensing timelines can stretch beyond two years – or longer, as many authorities struggle to recruit and keep talent with real experience in both banking and payments, and in technology.

Stablecoins sit at the intersection of securities law, payments regulation, banking supervision, and cutting‑edge cryptography. When regulators have limited opportunities for staff to be seconded into actual payment operations, or lose their qualified people to the private sector, the result is often cautious, reactive policymaking.

The good news for the industry as that authorities in hubs like Hong Kong and the UAE see licensing stablecoin operators as vital to their ambitions to lead in digital assets. They have either published comprehensive rules, or have established principles-based frameworks to focus on KYC and AML outcomes.

But central banks lack an established body for networking and trading policy ideas, so harmonization is a long, painful process. The closest thing to a coordinating body is the Bank of International Settlements, which sponsored mBridge – which, when it launched, caused a furor in the US, which saw it as BIS enabling China to work outside the framework of SWIFT-enabled correspondent banking. Don’t expect the BIS to be an active coordinator for stablecoins.

Transparency

Transparency is the another sticking point. This one is a head-scratcher for people in the crypto industry. They live with a kind of raw transparency, in which every transaction is visible. It’s the wallet addresses that are cryptographically hidden. But the kind of visibility available in crypto does not map to TradFi, which means transparency around stablecoins has to be shoehorned into legacy infrastructure.

In the SWIFT world, banks can see each counterparty and maintain audit trails for who owns which account at every hop. With stablecoin flows, the bank only sees a transfer from or to a crypto exchange or wallet provider, and cannot readily map the full chain of wallet owners touching the asset. That opacity clashes head‑on with AML, KYC, sanctions, and fraud obligations, especially in an era when billions of dollars have been lost to cybercrime and where blockchain transactions are irreversible.​

When a Visa or Mastercard payment goes wrong, the card network can reverse the transaction and often eat the fraud cost (hence justifying high issuer fees); on a public blockchain, there is no global dispute‑resolution layer.

Regulators and compliance teams are right to be wary: faster, irreversible money is great if you trust every actor in the system and can identify them; it is terrifying if you cannot.​ Perhaps there is a world in which TradFi could embrace crypto’s market transparency as a massive efficiency gain, but the more likely outcome is that stablecoins have to be designed around traditional compliance.

Privacy and identity

What might such a solution look like, one that enables the widespread adoption of stablecoins as a superior mode of payment, while preserving both the privacy demanded by crypto natives and the traditional safeguards against crime?

Today, many stablecoin systems rely on denial lists (blocking known bad actors) whereas banks operate on white lists, where only pre‑vetted clients can transact. Bridging that gap requires a new layer of decentralized identity and permissioned privacy that lets multiple participants, not just the issuer, onboard and vouch for clients.​

For banks and regulators, the core question is: who is this wallet? If a central securities depository, a custodian, or a bank can attach a verified identity to an on‑chain address using robust KYC/KYB processes, then stablecoin transfers can be treated much more like account‑to‑account transfers in today’s system. Technologies like zero‑knowledge proofs hold out the promise of allowing parties to demonstrate compliance with certain rules (for example, that a user is not on a sanctions list, or that a transaction is below a threshold) without revealing all underlying personal data.​

Some central‑bank digital currency experiments hint at this direction. In places like Hong Kong and mainland China, pilots such as eHKD and eCNY have been used to explore what happens when identity frameworks like iAMSmart or national ID systems are intertwined with digital money. Those projects raise uncomfortable questions about authoritarian overreach and surveillance, but they also provide technical templates for embedding identity at the protocol layer. For stablecoins, the challenge is to borrow the good parts – strong identity, clear liability, auditable records – while avoiding the worst: centralized, granular tracking of every transaction.​

Compliance is the price of admission; without it, no major bank or regulator will allow stablecoins to carry significant flows. Identity and privacy standards are the negotiation: can we use technology to prove that the industry can know enough about its users to satisfy AML and systemic‑risk concerns, while avoiding building panopticons?

This is a hard nut to crack, but it is achievable. Remember, it’s the plumbing, not the peg, that matters most. Our final article in this series will look at other aspects to stablecoin plumbing that still need attention: liquidity and resilience.