The Great Re-Platforming: How Tokenisation, Stablecoins and AI Agents are rebuilding financial markets from the rails up introduction
By Greg Nunan (pictured), Senior Financial Advisor Private Wealth at Gallagher
A market built for a world that no longer exists
Modern financial market infrastructure — clearing houses, custodians, settlement cycles, exchange hours, correspondent banking — was designed for a world where the fastest thing in the system was a human being with a phone. Trades settled in T+2 because paperwork and reconciliation took that long. Markets closed at 4pm because traders needed to sleep. A $5 payment wasn’t worth processing because the cost of moving it exceeded its value.
Almost none of those constraints exist today. What’s replacing them isn’t a single innovation — it’s the simultaneous, mutually reinforcing arrival of three things: tokenisation (representing ownership of an asset as a digitally native, programmable token rather than an entry in a proprietary database), stablecoins (dollar-denominated tokens that settle instantly, 24/7, without a bank in the middle), and AI agents (software that can now hold a wallet, make decisions, and transact autonomously). Individually, each is interesting. Together, they are starting to look like a genuine re-platforming of financial market infrastructure — not an alternative to it.
This essay works through seven concrete, currently-live developments that illustrate the shift, then turns to the uncomfortable question for anyone working in Australian financial services: what happens if we don’t move with it.
The new customer is a machine, and it transacts at a scale humans never could
For as long as finance has existed, the assumption baked into every payment rail — cards, ACH, SWIFT, BPAY — has been that a human is on at least one end of the transaction, authorising it, and that the transaction is large enough to justify the fixed cost of processing it. Visa and Mastercard interchange, card scheme fees, and correspondent banking spreads all assume a floor transaction size well above a dollar.
AI agents break that assumption completely. An autonomous agent calling a pricing API, retrieving a data feed, invoking another agent’s tool, or paying for a slice of GPU compute doesn’t need a human in the loop, and the transaction it wants to make might be worth a tenth of a cent. There was, until recently, no rail on earth that could process that economically.
That’s the gap the x402 protocol was built to close. Originally developed and open-sourced by Coinbase, x402 revives the dormant HTTP 402 “Payment Required” status code from the early web and turns it into a functioning payment rail: a server asks for payment, an agent’s wallet signs a stablecoin transfer (typically USDC), a facilitator verifies and settles it on-chain, and the resource is released — all within seconds, for a fraction of a cent, without an invoice, a card number, or a human approval step. In April 2026, Coinbase handed governance of the protocol to the Linux Foundation, giving it a neutral, non-corporate home — a strong signal that this is being built as public infrastructure, not a proprietary moat. Stripe and Tempo have launched a parallel, complementary standard, the Machine Payments Protocol (MPP), aimed at the broader set of machine-to-machine flows: recurring billing, subscriptions, and multi-agent settlement rather than just pay-per-request. Visa has built support for the protocol into its Trusted Agent Protocol, and Amazon has shipped agent payment tooling directly into Bedrock, its cloud AI platform — meaning a developer can give an agent a stablecoin wallet and spending controls with a few lines of code, no crypto expertise required.
The volumes are still early but the trajectory is the point: by April 2026, independent tracking put x402 activity at roughly 165 million agent transactions and around $50 million in cumulative volume across some 69,000 active agents — up from essentially zero eighteen months earlier. CoinGecko now prices some of its API access at a flat $0.01 per request over x402, priced explicitly for machine buyers rather than human ones.
Why does this matter beyond the technical novelty? Because it creates an entirely new category of demand for stablecoins and blockchain rails that has nothing to do with speculation. Every AI agent that needs to pay for data, compute, or another agent’s services is a recurring, structural source of on-chain transaction volume — and unlike a human trader, it never sleeps, never takes a public holiday, and scales with the number of agents deployed, not the number of humans employed.
Everything tradeable is becoming a 24/7 token
Robinhood‘s evolution over the past eighteen months is the clearest single case study of where public markets are heading. In June 2025 the company switched on tokenised versions of more than 200 US stocks and ETFs for European users, trading on Arbitrum inside its existing crypto app. By early 2026 that catalogue had grown past 2,000 tokens. On 1 July 2026, Robinhood went further, launching the public mainnet of Robinhood Chain — its own purpose-built Ethereum Layer-2 blockchain, engineered specifically for tokenised real-world assets, DeFi applications, and, notably, autonomous AI-agent trading. Tokenised stock trading went live the same day across more than 120 countries, running genuinely 24/7 rather than during exchange hours, alongside a DeFi lending product (“Robinhood Earn”) offering roughly 7% yield on its USDG stablecoin.
The regulatory nuance here matters and is worth understanding rather than glossing over. Robinhood’s global Stock Tokens are structured as debt securities referencing the underlying share held in custody by a US broker-dealer — the token-holder owns a claim, not the share itself, and has no shareholder rights (voting, for instance). Coinbase has taken a structurally different approach, offering tokenised US equities backed one-for-one by the actual underlying shares, with automatic on-chain dividend payment — closer to a genuine ownership wrapper. The SEC issued guidance in January 2026 explicitly distinguishing issuer-sponsored tokenised securities (which can represent true ownership) from third-party synthetic products (which face heavier scrutiny) — a sign that regulators are already trying to draw lines around a market that’s moving faster than the rulebook.
The scale is still modest in absolute terms — the entire tokenised real-world-asset market sat at roughly $30–38 billion as of mid-2026, with tokenised stocks specifically crossing the $1 billion mark in the second half of 2025 after 128% growth in that half alone. But scale isn’t really the story yet; architecture is. Nasdaq has formally petitioned US regulators to allow tokenised trading of listed equities on its own exchange. Once the largest, most conservative pieces of market infrastructure start asking permission to tokenise, the direction of travel stops being a crypto-industry talking point and becomes a market-structure inevitability.
There is a genuine risk sitting inside this growth, and it’s worth naming rather than glossing over for balance: Robinhood Chain’s early volumes were dominated not by tokenised real-world assets but by memecoin speculation, because a permissionless, cheap-to-use chain reliably attracts speculative trading faster than institutional adoption. Robinhood‘s own management has effectively set itself a public test — if tokenised real-world assets grow well beyond their early base while speculative activity fades, the strategy is working; if real-world assets stay flat while speculation moves on to the next chain, the infrastructure will have attracted a wave of hype without ever becoming the thing it was built to be. That test is still running.
The world’s largest asset managers are re-platforming the plumbing, not just launching a product
It would be easy to dismiss tokenisation as a retail-brokerage phenomenon if the world’s largest fund managers weren’t moving in exactly the same direction, for exactly the reasons a traditional asset manager would care about: settlement speed, collateral efficiency, and operational cost.
Franklin Templeton has arguably gone furthest. Its Franklin OnChain US Government Money Fund (“BENJI”) is one of the first US-registered funds to record share ownership natively on a public blockchain. In August 2026, the firm secured what it describes as the first SEC clearance allowing conventional, plain-vanilla ETFs and mutual funds to hold or use BENJI as collateral — meaning an ordinary investor in an otherwise conventional Franklin fund could, from as early as Q4 2026, end up holding a tokenised asset inside their portfolio without ever having sought crypto exposure at all. Franklin’s head of digital assets, Sandy Kaul, has framed the motivation in explicitly operational terms: more precise fund management, better yield capture, and tighter control over cash liquidity buffers — not ideology, not speculation, just better plumbing.
BlackRock, the world’s largest asset manager, is reportedly exploring its own tokenised ETF structures on public blockchains, building on the success of its BUIDL tokenised money-market fund (which has grown past $2 billion since its 2024 launch and gained real traction as collateral on crypto trading platforms). Larry Fink has publicly framed tokenisation as a mechanism to widen access to private markets — an unusual alignment between a firm managing around $10 trillion in client assets and infrastructure that started life as a niche crypto-native experiment.
The pattern across both firms is the same: tokenisation is entering traditional finance through the back door of operational efficiency, not the front door of investor demand for crypto exposure. That’s a more durable form of adoption than a speculative retail wave, because it doesn’t depend on continued enthusiasm — it depends on whether the plumbing is actually cheaper and faster, which, for short-duration cash instruments at least, it appears to be.
Consumer credit is being compressed by the same infrastructure
The tokenisation story usually gets told through capital markets, but the same underlying shift — moving underwriting, verification and settlement onto faster, more automated rails — is already reshaping consumer lending. US lender Figure has built a blockchain-based home equity lending product offering five-minute approval and funding in as few as five business days, against an industry average of roughly 39 days from application to approval and a typical 2–6 week close-to-funding timeline for a conventional home equity loan or HELOC. Figure has originated more than $25 billion in home equity products using this infrastructure, relying on automated valuation models instead of in-person appraisals and a digital, blockchain-recorded underwriting trail to remove most of the manual steps that make conventional home-equity lending slow.
This matters for the broader thesis for two reasons. First, it demonstrates that the efficiency gains from these rails aren’t confined to institutional capital markets — they compress cycle times in ordinary consumer finance too, wherever a process was previously slow because of manual verification and reconciliation rather than genuine complexity. Second, it’s a preview of what happens as the same infrastructure gets applied to mortgage, lending and even superannuation processes in other jurisdictions — including, eventually, Australia.
Coinbase is assembling a genuine bank-challenger, deliberately
Coinbase‘s evolution deserves to be read as a coherent strategy rather than a series of unrelated announcements. In April 2026, the OCC granted Coinbase conditional approval for a national trust bank charter — a move the Independent Community Bankers of America publicly called “a grave mistake”, precisely because it could give a crypto-native firm access to federal banking infrastructure without the same capital and reserve requirements as a chartered commercial bank. Coinbase has been explicit that it isn’t launching a deposit-taking, fractional-reserve bank — but the charter gives it federal regulatory uniformity for custody, and a foundation to build out payments infrastructure.
Around the same time, Coinbase’s own CEO Brian Armstrong disclosed that the exchange holds roughly $420 billion in customer assets — enough, measured the way a bank measures itself, to make it the 21st-largest bank in the United States by that metric, or the 8th-largest brokerage by assets under management. Armstrong has said a senior executive at one of the world’s ten largest banks told him crypto is now that bank’s “existential” issue — not a peripheral risk, but the central one.
Most recently, Coinbase has partnered with payments infrastructure firm Moov to bring stablecoin rails to more than 1,000 smaller US community banks and credit unions — a move that simultaneously builds Coinbase’s distribution and gives smaller banks a way to compete with the stablecoin capabilities megabanks are racing to build themselves. CB Insights’ partnership data shows Coinbase, alongside Ripple and Circle, forging more relationships with the traditional banking system than any other crypto-native firms in 2025 — not to disrupt banks from outside, but to become embedded inside the infrastructure banks themselves rely on.
Put together: Coinbase isn’t trying to become a bank in the traditional sense. It’s trying to become the infrastructure layer banks increasingly depend on, while simultaneously accumulating enough customer assets and payment volume that the distinction between “exchange”, “bank”, and “payments company” stops being meaningful.
Meme coin culture has become a genuine, chaotic, parallel funding market
It’s tempting — and often correct — to dismiss meme coin trading as pure speculation with no lasting economic function. But the capital-formation numbers now rival what venture capital raises for genuine early-stage companies, and that’s worth taking seriously as a market-structure phenomenon even where the underlying assets themselves have little intrinsic value.
Pump.fun, the dominant Solana-based memecoin launchpad, has facilitated the creation of more than 11 million tokens since its January 2024 launch and generated upward of $780 million in cumulative protocol revenue. Its own token sale — a public raise completed in roughly 12 minutes — brought in $600 million, on top of a $720 million private round before it, implying a valuation in the billions for a platform whose core product is letting anyone launch a speculative token with no vetting. Newer entrants like LetsBonk have taken meaningful market share by adding anti-sniper protections and creator revenue-sharing, suggesting the sector is professionalising rather than simply repeating its 2024 cycle. $PONS on the Robinhood chain and brokerage has taken this to a new level in recent weeks.
The comparison to venture capital isn’t a stretch. A traditional seed or Series A round takes months of due diligence, term-sheet negotiation, and legal structuring to raise a fraction of what a successful token launch raises in minutes, from a far larger and more geographically distributed pool of capital, with essentially no vetting process at all. That’s not a healthier way to allocate capital — the failure rate and fraud risk are enormous, and most tokens launched this way go to zero — but it is a faster one, and speed of capital formation is exactly the kind of thing that erodes an incumbent’s position over time even when the incumbent’s process is objectively more prudent. Venture capital’s moat has always rested partly on being the only efficient way to aggregate large pools of speculative early-stage capital; that moat is now visibly under pressure from a permissionless alternative operating at a completely different clock speed.
Stablecoins are becoming a genuine US Treasury-market story
The final piece connects all of the above back to the most traditional corner of finance there is: government debt. Under the GENIUS Act, signed into law in July 2025, US-regulated stablecoin issuers must back their tokens one-for-one with cash or highly liquid assets — in practice, overwhelmingly short-dated US Treasury bills with 93 days or less to maturity. That single design choice creates a direct, mechanical link between stablecoin growth and Treasury bill demand.
The numbers underline how quickly this could scale. The US Treasury’s own advisory committee has estimated stablecoin issuers already hold more than $120 billion in T-bills, and has modelled a scenario in which that could approach $1 trillion by 2028. Treasury Secretary Scott Bessent has publicly floated a stablecoin market exceeding $2 trillion by 2028, and as much as $3 trillion by 2030 — figures that, if realised, would make stablecoin issuers among the largest holders of short-dated US government debt in the world, alongside money-market funds and foreign central banks. ProShares has already launched a dedicated money-market ETF (IQMM) structured specifically around the GENIUS Act’s 93-day Treasury-bill requirement, effectively building a product to sit on the supply side of this new demand.
This is the part of the story that should reframe how traditional finance thinks about crypto altogether. Stablecoins are not simply a payments innovation happening in a corner of the market disconnected from “real” finance — they are becoming a structural, and growing, source of demand for the debt instrument that underpins the entire global financial system. The GENIUS Act didn’t just legitimise stablecoins; it wired them directly into the plumbing of US fiscal policy.
What happens if Australia doesn’t move
Every one of the seven shifts above is happening in a jurisdiction — overwhelmingly the United States — that has, for all its regulatory messiness, chosen to legislate and license rather than wait. Australia has not yet made that choice at the same pace, and the gap is measurable rather than theoretical.
Australia is, by several measures, one of the most crypto-engaged populations in the world: adoption is estimated at around 31% of the population, and crypto holdings inside self-managed superannuation funds have grown to roughly A$1.7 billion. That’s a large, motivated, already-engaged client base sitting inside a regulatory framework that is only now catching up.
The legislative response — the Corporations Amendment (Digital Assets Framework) Act 2026 — received Royal Assent in April 2026 but doesn’t fully commence until 9 April 2027, after an 18-month transition period. In the interim, ASIC’s temporary no-action position under Information Sheet 225 expired on 30 June 2026, after which firms operating without the correct AFS licence authorisation face penalties of up to 10% of annual turnover. The new regime introduces two genuinely new regulated categories — Digital Asset Platforms and Tokenised Custody Platforms — both requiring AFSL authorisation, shifting the regulatory question from “is this token a financial product?” to “does this platform’s structure and role trigger financial services obligations?”. It’s a more coherent framework than what existed before, but it arrives roughly two years after the SEC cleared Franklin Templeton to fold tokenised assets into mainstream US funds, and well over eighteen months after Robinhood, Coinbase and Nasdaq had already restructured core parts of their businesses around this shift.
ASIC’s own leadership has been unusually candid about the risk this creates. Chair Joe Longo has warned publicly that Australia risks becoming a “land of missed opportunity” if it fails to adapt to blockchain-driven change, including tokenisation — a striking admission from a regulator, and one that should be read as a signal to the industry as much as to government. ASIC’s Key Issues Outlook for 2026 separately flagged digital assets and AI-driven financial services as priority areas for supervisory attention, precisely because rapid innovation by participants unfamiliar with financial services law is creating both consumer risk and competitive distortion.
The practical risk for Australian advisers, licensees and platforms is not abstract. If US-based and offshore providers arrive in the Australian market with tokenised, 24/7, AI-native cash management, fixed-income access, and portfolio-efficiency products before local infrastructure and licensing genuinely catches up, the categories most exposed are the ones traditional wealth management has always considered its core turf: cash and short-duration fixed income (now competing with tokenised money-market products offering equivalent yield with better liquidity), portfolio administration and rebalancing (increasingly automatable by AI agents operating at machine speed across tokenised holdings), and even basic advice relationships (as clients start asking why a global, always-on, tokenised alternative isn’t already part of their portfolio). None of this requires Australian regulation to fail — it only requires the rest of the world to move first while local licensees wait for certainty that has, by global standards, already arrived elsewhere.
Where this leaves advisers and licensees
None of this is a case for rushing client money into speculative tokens, memecoins, or unlicensed offshore platforms — if anything, the regulatory gap is precisely why caution with clients’ capital remains essential. But there’s an important distinction between prudence with client money and fluency with the mechanics reshaping the market those clients’ money sits inside. Understanding how x402 and stablecoin rails actually work, how tokenised money-market funds function as collateral, why GENIUS Act stablecoins are becoming a Treasury-market factor, and what regulatory categories the Digital Assets Framework Act actually creates, is rapidly becoming table stakes for anyone advising on portfolios, cash management, or fixed income — not a specialist crypto skill set sitting off to the side of “real” advice.
The firms and advisers who build that fluency now, ahead of full DAF Act commencement in April 2027, won’t just be positioned to answer client questions when they arrive. They’ll be positioned to have shaped how those products are actually adopted in Australia, rather than reacting to infrastructure that was designed somewhere else, for someone else’s market, first.
This is general commentary on market structure and technology trends, not financial product advice.
