That’s what a lot of DeFi builders miss. They look at lending and follow models used elsewhere, pool capital, match borrowers and then just market the yield.

For short-term loans and liquid collateral, that’s often enough. But for multi-year real estate lending? It doesn’t scale. Because lending at scale is less about matching and more about infrastructure.

I did this for years as a broker and it’s obvious where the friction lives. It’s the machinery around lending: documents, checks, handoffs and internal systems that don’t talk to each other. You spend more time chasing a valuation, an underwriter query or a case update than you do thinking about credit.

And here’s the bit that most people miss, banks and lenders don’t keep lending because they’ve got endless spare cash. They keep lending because they can recycle capital.

They originate loans, hold them briefly, then securitise - package those loans and sell them to long-duration capital (pension funds, insurers, asset managers). That’s how they free up the balance sheet to do the next round.

Without securitisation, lending growth is capped by deposit growth and how much balance sheet they can raise. Simple as that.

Peer-to-peer lending has the opposite problem, it will need a constant and voluntary stream of capital willing to lock up for 5, 10, 30 years. You could attract liquidity for a while but you can’t reliably sustain it at the national mortgage scale. Eventually the pool goes quiet. It took unique circumstances, but we saw this in the UK in 2020 when securitisation markets basically shut during the first lockdown, the non-bank mortgage lenders (the ones without deposits or Bank of England backstops) were the ones that got squeezed, many non-bank lenders had to stop new lending.

Then the moment funding windows reopened, you saw issuance come back and lending capacity follow, it was someone like Kensington who then raised £650m+ through wholesale markets and talked openly about markets reopening after lockdown.

This is why the plumbing matters. Securitisation today is slow for one core reason: there’s no single source of truth.

Origination data in one system. Servicing in another. Accounting somewhere else. Nothing cleanly reconciles in real time. So every securitisation turns into a mini archaeology project, at great expense.

A bank reaches the point where they want to package a mortgage pool and sell it so first they pull data from multiple internal systems, then they reconcile discrepancies, then they build data rooms, then they pay third parties to verify samples, then rating agencies and lawyers do their part, then the sale gets arranged. Months later, capital is freed up and the cycle can restart.

It’s obviously down to legacy systems, they were built to do one job each, not to stay perfectly consistent across the entire loan lifecycle and the results are not great. Some of the most sophisticated institutions on earth can be unable to answer basic questions about their loan book without manual work. But over time, the centre of gravity will shift and not through a big-bang migration but because the clean book is simply easier to fund, audit and securitise.

So if this is hard for plain-vanilla mortgages, ask yourself what happens when you bolt tokenized collateral on top. You are no longer just proving the loan data is clean, you’re proving the asset rail is clean too. Where does the token live? Who validated the underlying property and valuation? What’s enforceable in the real world and not just code?

It’s the same messy back office, plus a whole new surface area for doubt.

People can approach this backwards, by tokenising assets first and only then persuade banks to lend against them, but I feel that if you want banks to adopt new rails you need to give them a reason that improves their current business alongside a theoretical future one.

The real wedge can be the mortgage book banks already have and the systems they run on. Put that on infrastructure that makes securitisation faster, cheaper and more auditable. Make the data consistent across origination, servicing and reporting. Make the audit trail native and not reconstructed. Then tokenized collateral actually just becomes an add-on rather than another new system.

Now imagine a bank’s existing mortgage book but sitting on private, permissioned rails where every payment is recorded automatically and every modification tracked. With a complete audit trail by default.

Not only are day-to-day or customer service enquiries far easier, when it’s time to securitise, there’s nothing to ‘compile’. The data is actually already there and more importantly it's instantly verifiable. Investors can see loan performance without waiting for a manual data room to be stitched together and due diligence becomes verification. That doesn’t remove the need for legal opinions, representations and warranties or jurisdiction-specific enforceability analysis: those don’t disappear. What changes is their scope. Instead of proving that loan data exists and is consistent, legal and rating work becomes focused on structure, risk and jurisdictional nuance rather than basic data reconstruction.

And once that infrastructure exists, adding tokenized collateral is trivial. Everything is the same, it’s just a new collateral type.

The major issue I see is that for this to work, it’s going to be a struggle to build it all themselves. I liken it to the shift from paper to computers: quite a large undertaking, years of work, specialist knowledge and by the time you finish, the tech has already moved. It is typically why they outsource the complexity and focus on what they actually do well: underwriting, customer relationships, credit decisions.

The same thing I expect to happen here, we are already seeing it in different parts of their business. Banks will use infrastructure that handles interoperability, privacy, auditability and all the ugly plumbing while they focus on business.

This is the opportunity most DeFi builders are missing. It’s not replacing banks with protocols and it’s not building infrastructure only for a tokenized future. We’ve already seen early versions of this work. Figure and others proved that digitised loan rails and on-chain securitisation are possible. But those pioneers were necessarily first movers, operating outside the core banking system. The next phase is about doing it better, at scale, with banks leading now the model is proven and the risk profile is understood

It’s fixing how banks lend today, against the collateral and loan books they already have, using rails that just happen to be compatible with tokenized assets too.

Because normal people borrowing against their home don’t care about blockchain. They care about getting approved quickly, getting a fair rate and dealing with an institution they recognise and trust.

If you can give banks infrastructure that makes their current lending machine faster and cleaner, while making securitisation feel modern then you’ve solved the real problem.

And when tokenized real estate needs financing at scale, it won’t require a new lending system, It’ll just be another collateral type on rails that already works.