The Things That Don't Exist Yet

Most people look onchain and see a faster, more efficient version of what we already have. Faster payments. Lower-cost settlement. More efficient capital markets. And they’re right. That alone is a massive opportunity, and it will produce many venture-scale outcomes over the next decade.

But I think it’s the smaller part of the story.

When I look at the technology, at what you can actually do with programmable assets in a global, composable, always-on environment, I think we’ve barely explored the surface area of what’s possible. The most incredible things haven’t been built yet. And the reason they haven’t been built is not that the technology isn’t ready. It’s that we haven’t finished imagining them.

The Email Trap

When the internet first arrived, the obvious use case was communication. Email was faster, cheaper letters. And email was a big deal, but it wasn’t built to make the post office faster. It was its own thing and it became ubiquitous. So if you were evaluating the internet in 1995 and saw email reaching mass adoption, you could have reasonably concluded that the thesis had played out.

But the bulk of the opportunity hadn’t even been conceived yet. Search. Social networks. E-commerce. Cloud computing. SaaS. Streaming. None of these were “faster versions of something that already existed.” They were net-new categories that could not have existed before the internet created the conditions for them. Google wasn’t a faster library. Facebook wasn’t a faster phone book. AWS wasn’t a faster server room. They were things that only made sense once you had a globally connected, programmable network.

Collectively, these net-new categories dwarfed the “faster communication” use case by orders of magnitude.

I think crypto is in its email moment right now. Most of the attention is focused on making existing financial products work better on onchain rails. Faster settlement. Cheaper cross-border payments. Tokenized versions of treasury bonds and equities. More efficient lending markets. And it's working too: stablecoins settled $33 trillion in 2025 and tokenized treasuries recently crossed $15 billion. The largest asset managers and banks in the world are building on public chains.

This is great. I’m excited about all of it. I invest in it every day. But it’s the obvious use case, the one that maps cleanly onto things we already understand, and it’s big enough that it’s easy to mistake it for the whole opportunity.

The question that engages me even more is: what are the things that only become possible when you have programmable assets in a global, composable, always-on, permissionless environment? What are the new verbs, the categories that don’t have names yet?

What A New Verb Looks Like

We have at least one clean example, and it’s worth walking through because it illustrates the kind of thing I think we’ll see a lot more of.

What if you could borrow a billion dollars with no collateral and the lender had a mathematical guarantee of repayment?

That’s flash loans: borrow any amount of capital, with zero collateral, so long as it’s returned within the same transaction. If it isn’t returned, the entire transaction reverts as if it never happened. No risk to the lender. No credit check. No relationship. No collateral. Just a guarantee enforced by the logic of the system itself.

Before flash loans existed, nobody asked for them. Why would they? The concept doesn’t map onto anything in traditional finance. It’s not even useful before the advent of programmable assets and so there’s no pre-existing category to improve. Borrowing without collateral, at unlimited size, with guaranteed repayment isn’t a feature you’d ever request because in any system where transactions take time, it’s impossible. It only becomes possible when execution is atomic, assets are programmable, and when an entire sequence of actions either completes in full or doesn’t happen at all.

Once atomicity made it viable, flash loans became a standard tool in the onchain economy, used for arbitrage, liquidations, collateral swaps, and capital efficiency strategies that have no analog on traditional rails. And yes, like any powerful new capability, they’ve been used adversarially too, which only reinforces how novel the underlying mechanism is.

Flash loans didn’t make borrowing faster or cheaper. They created a type of borrowing that was structurally impossible before programmable assets and atomic execution created the conditions for it. That’s what I mean by a new verb. A new action. Something the system can do now that it couldn’t do before, not because someone found an optimization, but because the primitives themselves changed.

Limits Of Imagination

But I have to be straightforward about the limits of my own imagination.

I can describe the design space in abstract terms. Public blockchains introduced a set of primitives that didn’t exist before: atomic execution, shared global state, programmable escrow, deterministic settlement, composability across independent actors, assets that are software. We’ve never had a financial system where settlement, custody, clearing, and enforcement all live inside the same programmable environment. When layers that used to be separate collapse into one, new things become possible.

But I can’t tell you exactly what most of those things are. And I think that’s actually the point.

Human imagination is backward-looking. We’re very good at taking what exists and making it better. We’re much worse at conceiving of things that were actually impossible yesterday. We look at onchain rails and instinctively ask: what existing product can this make faster and cheaper? The harder and more valuable question is: what can this create that has no predecessor?

I have intuitions. Programmable escrow that enforces complex agreements without intermediaries. Capital that delegates itself to software agents operating within bounded constraints. Financial structures that assemble and dissolve in real time based on conditions verified onchain. These feel directionally right. But the most important applications are probably ones I can’t describe yet, because they don’t resemble anything I’ve seen before.

The inability to enumerate them is actually the strongest version of the thesis: if I could easily list all the net-new things, they wouldn't really be net-new. The design space is vast, mostly unexplored, and too large for intuition alone to map. Again, that's the point.

As a result, most attempts to build in this space will fail. The design space being large doesn't mean the outcomes are easy. But the opportunity for the things that do work is extraordinary, and identifying them early, before they're obvious, is what we've spent the last thirteen years building the pattern recognition to understand. And it's the part of the opportunity that gets me most excited about the decade ahead.

Most Of The Opportunity is Ahead

If the internet analogy holds, then the equivalent of search, social, cloud, and SaaS for the onchain economy hasn’t been built yet. Email was a trillion-dollar category. The things that came after it were worth tens of trillions.

I think we’ll look back in ten years and the things we’re most excited about will be things that don’t exist today. Not just efficiency improvements to banks or exchanges or asset managers. Things that only make sense once you have programmable assets in a composable, global, always-on environment. Things that will feel obvious in hindsight but that we can’t quite see yet because they don’t have predecessors to point to.

Flash loans gave us one glimpse, but there will be many more. The design space is enormous and we’re early in exploring it.