Infrastructure, Not Products

Why the banks defining the next era of finance aren't launching tokenization pilots, they're rebuilding their pipes.

Apr 19, 2026

In the last few weeks, the SEC approved Nasdaq’s framework for trading tokenized stocks on blockchain rails. NYSE announced a partnership with Securitize to build a 24/7 digital trading platform for tokenized equities and fixed income. DTCC is building an MVP to tokenize DTC custodied Treasury securities on the Canton Network. SWIFT completed a multi-bank tokenized bond settlement trial. Goldman Sachs is spinning out its tokenization platform as an independent, industry owned company.

These are not pilot programs. These are the institutions that run the plumbing of global finance, and they are rebuilding that plumbing on programmable rails.

I think this is the most important thing happening in banking right now, and I think most bank executives are missing it. Because the conversation inside most banks is still about products: should we tokenize bonds? Launch a stablecoin? Offer tokenized deposits? Those are reasonable questions. But they’re the wrong starting point. The starting point should be infrastructure.

The product trap

TRAPS - Definition & Meaning - Reverso English Dictionary

Anyone who approaches this as ‘which products are suitable for tokenization’ is likely to make a first principle error. I think the right framing is whether blockchain is a baseline infrastructure layer that cuts across the organization. Not a product. A platform.

When FIs treat tokenization as a product question, the pattern is looks very predictable. Someone in the innovation team runs a pilot. They tokenize a bond, or a repo transaction, or a fund share. The pilot works. There’s a press release. And then not much happens, because the pilot was a product, and products need infrastructure to scale, and the infrastructure doesn’t exist inside the institution yet. So the pilot sits in a sandbox, demonstrating what’s possible without changing how the FI actually operates.

That distinction matters for how FIs allocate capital, structure teams, and think about competitive positioning. Pains me to say this as proud product person, but we have to guard against product thinking leading to small budgets in innovation labs, isolated from the core business. If we have the end in mind, then infrastructure thinking leads to multi-year investment in core systems, organizational redesign, and strategic partnerships. One produces demos. The other produces capability.

McKinsey’s base case for tokenized financial assets (excluding stablecoins and CBDCs) is roughly $2 trillion by 2030. BCG’s estimate, using a broader scope that includes tokenized money, is $16 trillion. The gap between those numbers isn’t an analytical disagreement. It’s the difference between counting tokenization products and counting the infrastructure those products run on. The infrastructure number is an order of magnitude larger because infrastructure creates network effects that individual products can’t.

What’s actually being built

The reason I’m writing this now is that the infrastructure layer is forming fast, and I don’t think most bank executives appreciate how much has moved in the last twelve months.

Start with the pipes. DTCC, the backbone of US securities settlement, announced in December 2025 that it will tokenize DTC custodied US Treasury securities on the Canton Network using its ComposerX platform. An MVP is targeted for the first half of 2026, with plans to expand to a broader range of DTC-eligible assets based on client demand. This isn’t a pilot. This is the entity that settles virtually every US equity and fixed-income trade building tokenization into its core infrastructure.

SWIFT, which connects over 11,000 institutions in more than 200 countries, completed a multi-bank tokenized bond settlement trial in January 2026 with BNP Paribas, Intesa Sanpaolo, and SG Forge. It wasn’t a proof of concept. It was the first time SWIFT orchestrated tokenized asset transactions as a single coordinated process using both blockchain and traditional systems. SWIFT is now adding a blockchain based ledger to its infrastructure, designed in collaboration with over 30 banks, initially focused on real-time, 24/7 cross-border payments.

The exchanges are moving too. On March 18, 2026, the SEC approved Nasdaq’s framework for trading tokenized stocks and ETFs on blockchain rails alongside traditional shares. Russell 1000 stocks and major ETFs can now trade and settle as blockchain-based tokens, with the same tickers, prices, and investor rights. Nasdaq tapped Kraken to distribute stock tokens internationally. Six days later, NYSE announced a partnership with Securitize to build a 24/7 digital trading platform for tokenized equities, ETFs, and fixed income, with a pilot targeted for Q3 2026.

Goldman Sachs is spinning out its tokenization platform, GS DAP, as an independent company by mid 2026. The reasoning is telling: having the platform on Goldman’s balance sheet made it hard to get broad institutional adoption. Other banks didn’t want to run their infrastructure on a competitor’s platform. Making it independent and industry-owned removes that barrier. Tradeweb became the first strategic partner, integrating its trading and liquidity capabilities across fixed income. GS DAP runs on the Canton Network, which now supports over $6 trillion in on-chain assets across 600 institutions.

BNY, the world’s largest custodian with nearly $58 trillion in assets under custody, launched tokenized deposits in early 2026 for collateral and margin workflows. But the bigger story is how Carolyn Weinberg described their approach: mapping every post-trade step end-to-end and asking what can be reimagined. Not “which product should we tokenize?” but “how does the entire post-trade process work differently on programmable rails?”

That’s infrastructure thinking. And it’s happening at the institutions that move markets.

The network economics

Here’s why I think the product vs infrastructure distinction matters so much economically.

Products compete on features. A tokenized bond competes with other bonds on yield, credit quality, and liquidity. If your tokenized bond is slightly better than the next one, you get a slightly bigger market share. The economics are linear.

Infrastructure creates network effects. A shared settlement layer becomes more valuable as more participants join. A common tokenization standard becomes more useful as more assets conform to it. A coordinated compliance framework becomes more efficient as more institutions adopt it. The economics are exponential. That’s a fundamentally different business case, and it requires a fundamentally different investment thesis.

Look at what’s forming. The Canton Network connects 600 institutions with built in privacy controls and cross-ledger interoperability. The Cari Network (KeyBank, Huntington, First Horizon, M&T, Old National) is building tokenized deposit infrastructure on ZKsync anchored to Ethereum. The IBAT consortium is creating a shared network for community banks in Texas. Custodia and Vantage’s tokenized deposit infrastructure is being adopted by 600 banks through the Participate network. These aren’t really product launches. They’re infrastructure buildouts where the value comes from the network, not any single node.

The analogy I find myself coming back to is Visa. In 1958, Bank of America launched BankAmericard as a product, a single bank’s credit card mailed to 65,000 people in Fresno. It was a product for eight years. Then in 1966, BofA started licensing it to other banks. By 1970, the issuing banks took over entirely, forming a member owned consortium that Bank of America no longer controlled. It was rebranded as Visa in 1976. The product became infrastructure, and the infrastructure became the most valuable payments network on earth. No single bank could have built that alone. But once the shared rails existed, every bank that plugged in could offer its own products on top: credit cards, debit cards, rewards, corporate cards. The products differentiated. The infrastructure was shared.

Goldman spinning out GS DAP follows the same logic. A single bank’s tokenization platform, made independent and industry owned so it can become shared infrastructure. The Cari Network, the IBAT consortium, all of them are similar pattern of banks recognizing that the value is in the network, not any single node.

The banks still asking which product to tokenize are thinking about the card. The ones building shared infrastructure are thinking about the network.

The interoperability imperative

This is where I think the infrastructure argument gets most urgent, because the biggest risk in the current buildout is fragmentation.

Right now, tokenized assets live on different blockchains with different standards, different compliance frameworks, and different settlement mechanisms. A tokenized Treasury on Canton can’t easily interact with a tokenized deposit on Ethereum, which can’t easily settle against a stablecoin on Solana. That fragmentation is fine for pilots and proofs of concept. It’s a disaster for infrastructure.

SWIFT seems to understand this. Its interoperability trials weren’t about building a single blockchain. They were about connecting different blockchains through a common coordination layer, so that a bank using one network can transact with a bank using another. The January 2026 trial demonstrated exactly this: seamless exchange and settlement of tokenized bonds across blockchain platforms and traditional systems.

Canton’s architecture works the same way. It’s designed as a “network of networks” where institutions operate independent but interoperable ledgers. Each institution keeps its own data private. The Global Synchronizer enables atomic transactions across applications without sacrificing confidentiality. That’s why DTCC chose it for Treasury tokenization: it preserves the privacy and control that regulated institutions require while enabling the composability that makes programmable settlement useful.

The lesson here is that infrastructure without interoperability is just a more expensive silo. Banks that build tokenization capability in isolation, on proprietary systems that don’t connect to the broader ecosystem, will find themselves with a product, not a platform. And in a networked world, products that don’t connect become dead ends.

Treasury as the proving ground

I’ve written before about how programmable settlement changes the economic character of deposits. But treasury is where infrastructure thinking meets real money in the most immediate way.

Think about what corporate treasury looks like today. Cash sits in bank accounts earning whatever the bank offers, or in money market funds that settle T+1 or T+2. Collateral is managed through manual processes with end-of-day reconciliation. Intraday liquidity is expensive and hard to optimize because the systems that manage it are slow. FX settlement still carries Herstatt risk for cross-timezone trades. Repo markets close on weekends.

Now think about what treasury looks like on programmable infrastructure. Cash is a token that can be moved instantly, 24/7, to wherever it generates the highest risk-adjusted return. Collateral management is automated through smart contracts that rebalance positions in real time based on predefined rules. Intraday liquidity is optimized continuously because the infrastructure never stops. FX settlement is atomic, eliminating counterparty risk. Repo can happen at 2am on a Sunday.

Someone at Alpha Point made this case directly at the recent American Banker On-Chain event

“This is not about incremental efficiency. This is a substantial change on the balance sheet.”

They’re right. This isn’t 5% cost savings on post trade processing. This is a structural change in how treasury generates and deploys capital.

The tokenized Treasury market is now over $11 billion, up from $3.9 billion at the start of 2025. Circle’s USYC has overtaken BlackRock’s BUIDL as the largest tokenized Treasury product at $2.2 billion. Goldman announced 24/7 tokenized Treasury bond trading. DTCC is tokenizing DTC custodied Treasuries. The infrastructure for programmable treasury isn’t theoretical. It’s being built, and the banks that connect to it first will have a structural advantage in how they manage their own balance sheets and serve their institutional clients.

What this means for bank boards

I want to be specific about what infrastructure thinking actually changes at the decision making level, because this is where I see the biggest disconnect.

When a bank treats tokenization as a product, the budget sits with a business line. It’s measured on product P&L. It competes for funding against other product initiatives. The team is small, often in an innovation lab, and the rest of the bank doesn’t need to change how it operates. The board hears an update once a quarter and nods politely.

When a bank treats tokenization as infrastructure, the investment is actually enterprise wide. It touches technology, operations, compliance, treasury, custody, and client servicing. It requires organizational change with either a dedicated unit like JPMorgan’s Kinexys or U.S. Bank’s Digital Assets and Money Movement group, or a consortium model like Cari Network, or a strategic partnership with an infrastructure provider like Stablecore. The board isn’t hearing an update. The board is making a multi-year capital allocation decision about the bank’s operating model.

That’s a harder conversation. It requires more conviction, more coordination, and more willingness to invest before the returns are fully visible. But it’s the right conversation, because the infrastructure being built right now will determine which banks are at the center of the financial system in 2030 and which are at the periphery.

Global banking IT spending is projected to exceed $760 billion in 2025, rising at 9% annually. Seven in ten banks plan to increase infrastructure spending. I don’t doubt that banks are willing to invest in technology. I think the challenge is directing that investment in the right layer. Another core banking modernization that preserves the existing architecture is a product investment. Connecting to the programmable settlement infrastructure that DTCC, SWIFT, the exchanges, and the major custodians are building is an infrastructure investment. One maintains the status quo. The other positions the bank for what’s coming.

The banks building pipes

I’ve been writing about programmable money for a while now, and the thing that strikes me most about where we are in early 2026 is how fast the conversation has shifted from “should we?” to “how do we?” The regulatory environment is favorable. The technology works. The business case is clear.

But the execution models that are succeeding aren’t the ones chasing tokenization products. They’re the ones building, or connecting to, infrastructure. BNY mapping post trade end to end. Goldman making GS DAP industry neutral. DTCC putting Treasuries on Canton. SWIFT adding a blockchain ledger. Five regional banks forming Cari Network. The pattern is infrastructure first, products on top.

Every major technology transition in banking has followed this sequence. ATMs started as a product (cash dispensing) and became infrastructure (a shared network that changed how banks distributed services). Electronic trading started as a product (screen based execution) and became infrastructure (the backbone of modern capital markets). Internet banking started as a product (check your balance online) and became infrastructure (the platform for everything from payments to lending to wealth management). In every case, the banks that built or connected to the infrastructure early captured disproportionate value. The ones that treated the new technology as just another product offering lost ground to those that understood the shift.

Programmable money is following the same pattern. The product phase (tokenize a bond, run a pilot, issue a press release) is likely winding down. The infrastructure phase (rebuild the pipes, connect to shared networks, redesign operations) will start accelerating.

The question for every bank board is simple really: are you building products, or are you building pipes? Because in five years, the answer to that question will be obvious. And for some banks, it will probably be too late to change it.

References

Tokenization market size and projections

DTCC and Canton Network

SWIFT tokenization and interoperability

Exchange infrastructure

Goldman Sachs GS DAP

BNY digital assets

Bank infrastructure consortiums

Banking IT spending

Atomic Settlement

Discussion about this post

Ready for more?