The Execution Gap

Why the challenge in programmable money isn't just the technology or the regulation, but the operating model

Apr 12, 2026

Large banks are 40 percent less productive than digital natives. Fintechs ship new product features every two to four weeks; traditional banks take four to six months. Taking a new product from concept to market takes a fintech three to six months. At a large bank, twelve to eighteen months. And roughly 70 percent of digital transformation initiatives still fail to meet their objectives.

Those aren’t statistics about programmable money. They’re statistics about banks trying to do anything new. And programmable money is harder than most things banks have tried, because it touches payments, custody, settlement, compliance, treasury, client servicing, and risk management all at once. It’s not a product launch. It’s an operating model change.

The industry consensus right now is that the remaining barriers to programmable money are regulatory and technological. I think that’s wrong. Regulation is the most favorable in memory. The GENIUS Act is law, signed in July 2025, with the OCC already issuing proposed rules to implement it. The OCC has published multiple interpretive letters through the year confirming that banks can custody digital assets, execute transactions, participate in blockchain networks, and pay gas fees. The Basel Committee published third-party risk principles in December 2025. The technology providers are lining up.

The real barrier is execution. The actual operational capability to run programmable money safely inside existing constraints. I’ve spent the last 25 years watching the gap between strategy and implementation kill good ideas in banking. Not because the strategy was wrong. Not because the technology didn’t work. Because the organization, its processes and systems and culture, couldn’t absorb the change fast enough to make it real. And I think programmable money may be the widest version of that gap we’ve seen.

File:Mind the gap 2.JPG - Wikimedia Commons

Where execution breaks down

Start with the systems. Over 75 percent of IT budgets at financial institutions go to maintaining what already exists. Celent’s 2025 data on corporate banking IT tells the same story: the vast majority of spend is mandatory, keeping the lights on or regulatory, and the share available for innovation is shrinking even as total IT spending rises.

The depth of the problem is easy to underestimate. Core banking systems are built on assumptions that programmable money violates. Batch processing cycles. End of day reconciliation. T+1 or T+2 settlement assumptions baked into ledger architecture. Currency fields that only accept three-letter ISO codes. At a recent conference, a bank exec from TD told a story about a corporate client asking how to define USDC in their ERP system. Nobody knew. The system wasn’t designed for a digital asset that isn’t a currency, isn’t a security, and doesn’t fit any existing classification.

That’s not a problem you solve with an API. It’s a data architecture problem, an accounting classification problem, and a process redesign problem, all rippling outward at once. How do you reconcile an onchain transaction with an offchain ledger? How do you account for a tokenized deposit that exists simultaneously as a bank liability and a blockchain token? How do you report it to regulators who haven’t standardized the format?

The banks making progress are mostly building alongside their legacy systems rather than trying to transform them. Sidecar architectures, middleware layers, separate digital asset platforms that connect back to the core through adapters. IDC projects that by 2028, 70 to 80 percent of banks will pursue sidecar strategies rather than core replacement. That’s pragmatic. It also tells you how deep the legacy problem runs.

Then there’s compliance. Banks in the US and Canada spend over $60 billion a year on AML compliance alone. The average bank allocates roughly $64 million annually to KYC and AML processes. Regulatory penalties for financial institutions totaled $3.8 billion in 2025. Every new activity, every new rail, every new asset type has to be integrated into a compliance infrastructure that is already stretched, already expensive, and already under scrutiny.

Now layer on programmable money. Stablecoins move 24/7 across borders in seconds. Tokenized deposits can transfer peer-to-peer on public blockchains. Smart contracts can execute transactions without anyone in the loop. Transaction monitoring needs to be real-time, not batch. Sanctions screening needs to happen before settlement, not after. Customer due diligence needs to account for wallet addresses, on-chain identities, and counterparties that may not exist in any traditional database.

Blockchain actually provides a better compliance substrate in certain respects. Every transaction is visible, timestamped, and immutable. Firms like Chainalysis and Elliptic can trace the entire history of a wallet. Programmable compliance, where sanctions screening and transfer restrictions are embedded directly in token contracts, is a growing capability. But getting there requires compliance staff who understand on-chain analytics, technology that can screen blockchain transactions in real time, and policies for questions that don’t have settled answers yet. What counts as a suspicious transaction on a blockchain? When does a wallet address trigger an OFAC obligation? How do you file a SAR for a smart contract interaction? This is solvable, and it’s being solved. But it isn’t free and it isn’t fast, and any bank that underestimates the compliance execution burden is going to have a bad conversation with their examiner.

The operational challenge that probably gets the least attention, though, is time. Programmable money doesn’t sleep. Blockchains run around the clock. Stablecoins settle at 3am on Sunday. Tokenized Treasury funds offer instant redemption day and night. I’ve written about how this changes the economic character of deposits. It also changes the operational character of a bank.

Banks are not built for always on. Settlement systems run during business hours. Treasury desks operate in shifts aligned with market hours. Risk monitoring has overnight gaps. IT maintenance windows assume downtime.

We already have a proof point for how hard this transition is. FedNow launched in July 2023 to enable real-time payments around the clock. Two years later, more than 1,500 financial institutions have joined, reaching about 40 percent of US demand deposit accounts, but still a fraction of roughly 10,000 eligible institutions. Most can only receive instant payments, not send them. Transaction volumes are growing fast ($245 billion in Q2 2025, up from $492 million a year earlier), but operational adoption is lagging well behind the technology.

And FedNow is just real-time payments. Programmable money goes further. If a bank offers tokenized deposits redeemable around the clock, it needs 24/7 liquidity management. If it participates in tokenized Treasury settlement, it needs 24/7 risk monitoring. If a corporate client’s smart contract triggers a payment at 2am Saturday, someone or something needs to be there to make sure it settles correctly, the AML screening happened, the liquidity was available, and the reconciliation is clean.

The T+1 settlement shift in May 2024 offered a preview. The industry prepared for months, and affirmation rates rose from 73 percent to 95 percent ahead of the deadline. But the preparation was enormous, and that was just compressing from T+2 to T+1, during business hours, for a well-understood asset class. Moving to around the clock settlement for tokenized assets is harder by an order of magnitude. The operational investment required is larger than most strategy presentations acknowledge.

All of this is downstream, though, of what I think is the binding constraint: talent. Sixty percent of Fortune 500 companies are now working on blockchain initiatives, according to Coinbase’s 2025 State of Crypto report. They’re all competing for the same small pool of people.

The gap isn’t blockchain developers. Banks can hire those, or partner with firms that have them. The gap is people who understand both worlds. People who can read a Solidity contract and a Basel III capital requirement and understand how they interact. Who can design a smart contract governance framework that satisfies the CTO and the chief risk officer. Who can translate between the language of DeFi and the language of bank regulation.

This talent barely exists because the two industries were built by people who chose opposite sides of a wall. Crypto attracted people who deliberately left traditional finance. Banking was built by people who have never deployed code to a blockchain. The intersection, where all the hardest execution decisions live, is tiny.

Rachel from Anchorage Digital made this point at the recent American Banker’s on-chain conference:

“If you want to move fast, partner with a digital native company and figure out where you want to build internally from there.”

That’s practical advice. But even with a partner, the bank needs internal capability to manage the partnership, evaluate the technology, understand the risks, and make governance decisions. You can outsource the engineering. You can’t outsource the judgment.

And then there’s the problem nobody wants to talk about: the org chart. Programmable money touches payments, treasury, custody, compliance, risk, technology, and client servicing simultaneously. In most banks, those are separate departments with separate budgets, separate leadership, separate incentives, and separate regulatory responsibilities. Getting them to work together on a shared initiative is a coordination problem that makes the technology look simple.

This is why U.S. Bank created a new organizational unit, Digital Assets and Money Movement, reporting to the chief digital officer. It’s why JPMorgan built Kinexys as a dedicated business unit with its own P&L spanning payments, assets, and data. It’s why Carolyn Weinberg at BNY described mapping every post-trade step end to end and reimagining it holistically, rather than letting each department optimize its piece.

Those are structural answers to a structural problem. But most banks, and especially mid-size and community banks, can’t create dedicated digital asset divisions. They have to execute within their existing structure, which means navigating crossvdepartmental politics, competing budget priorities, and leaders who see programmable money as someone else’s problem.

Anyone who approaches this as ‘which products are suitable for tokenization’ is likely to make a first principle error. The right framing is infrastructure, not product. But infrastructure projects require enterprise wide buy in. They require an executive sponsor who can force coordination across departments. They require, to be direct, power.

What’s actually working

I’ve painted a bit of a rough picture, and I want to be clear that banks are executing. Dozens of institutions are live or launching in 2026. This less about whether execution is possible, it’s more about which models work.

From what I can see, three are producing results.

The first is the dedicated unit. JPMorgan’s Kinexys, U.S. Bank’s Digital Assets and Money Movement organization, BNY’s Digital Asset Platform. These create protected space with dedicated leadership, budget, and talent, while maintaining integration points back to the core. This works for large institutions that can afford it. It doesn’t scale down.

The second is the consortium. The Cari Network (KeyBank, Huntington, First Horizon, M&T, Old National), the IBAT community bank consortium, the Texas Bankers Association pilot program, Custodia and Vantage’s integration with the 600 bank Participate network. These pool the execution burden across institutions, sharing infrastructure, compliance frameworks, and operational learning. This is particularly promising for community banks because it gives them access to capabilities they could never build alone.

The third is the partner model. Stablecore pre-integrating digital asset rails into core banking platforms. Anchorage offering white label stablecoin issuance. IO Builders providing tokenization middleware. TruStage building a credit union stablecoin. These let banks adopt programmable money without building deep internal engineering capability. They treat blockchain infrastructure the way banks already treat card processing or payments networks: as something they consume, not build.

Each approach has trade offs. Dedicated units require real investment and can become islands. Consortiums require governance across institutions with different priorities. Partner models create dependencies on third parties, which is exactly what the Basel Committee’s December 2025 principles were designed to address.

But they share something: they scope the problem to something manageable. None of them try to transform the entire bank at once. They pick a use case, build the operational muscle, and expand from there. Crawl, walk, run.

What this means

I’ve been arguing that programmable settlement is poised to reshape bank economics in ways that are structural, not cyclical. The settlement friction subsidy is eroding. Governance over programmable money is hard but tractable. Banks are moving, and the regulatory environment is supportive.

But here’s the thing about execution gaps: they don’t just slow things down. They sort the market. The banks that build operational capability early will be the ones that own the infrastructure layer for programmable money. The ones that wait for perfect conditions will find themselves consuming someone else’s rails, paying someone else’s fees, and depending on someone else’s compliance stack.

That’s the pattern we saw with card processing, with payments networks, with cloud infrastructure. The institutions that built operational capability around the infrastructure became the platforms. Everyone else became the customers.

The programmable money execution gap isn’t just a risk to manage. It’s the mechanism by which the next layer of financial infrastructure will get divided up. The banks that solve the execution problem won’t just survive the transition. They’ll own a piece of the plumbing. The ones that don’t will rent it.

That’s really not a technology question. It’s a power question. And I’m sure it’s being answered right now, by whoever is building operational capability while everyone else is still refining their strategy decks.

References

Bank digital transformation and productivity

Legacy systems and IT spending

Compliance costs and regulatory fines

24/7 settlement and real-time payments

T+1 settlement transition

Blockchain talent and hiring

ERP and treasury integration

Institutional execution models

Atomic Settlement

Discussion about this post

Ready for more?