Compute for Equity

The arbitrage I couldn't unsee

In late 2025, I watched a founder I respect raise a few million dollars and hand most of it straight to a cloud provider for GPUs. He was thrilled — and he should have been; he'd secured the compute his company needed to exist. But I kept doing the math on the way home. He had just sold a slice of his company for cash, then immediately spent that cash renting hardware. The bank, the round, the dilution — all of it was friction in the middle of a trade that, at its core, was simple: he had ownership, the cloud had compute, and they wanted to exchange them.

The arbitrage was obvious once I saw it, and then I couldn't unsee it. Compute is sold at a 5–10× markup over its marginal cost of energy and depreciation. A founder pays that markup in cash he raised dilutively. Meanwhile, the people who actually own the inputs — energy producers with surplus power, datacenters with idle GPUs — sit on capacity with no way to capture the upside of the intelligence it produces. Three of the most valuable assets of the AI era — energy, compute and equity — are each enormous, and each is priced in its own silo, unable to settle directly against the others.

That's not a capital problem. The capital exists; there is an estimated $490B of demand to build AI infrastructure, and a GPU-as-collateral lending market that went from zero to over $10B in under two years. It's a coordination problem. The markets that should connect are kept apart because there is no neutral institution to price both sides, become the counterparty to each, and clear the trade. Every time financial history has faced that exact gap — in commodities, in securities, in energy — the answer has been the same: a clearing house.

Why I'm building it, and why here

I'm a builder of products, not a banker. I've spent over a decade shipping consumer and AI products, and what I've learned is that the hardest markets aren't unlocked by a clever idea — they're unlocked by an institution patient enough to earn trust. So I'm building the clearing layer for the AI economy: a regulated venue where a startup can pay for compute with a tokenized Compute-SAFE, a datacenter can earn equity for capacity, and an investor can hold an asset-backed AI position — all priced by a cross-asset oracle and settled under real law.

It can only be built in Abu Dhabi first. The ADGM runs the world's most advanced framework for tokenized securities under English common law; the FSRA RegLab gives a clear path from sandbox to full license; the emirate pairs the cheapest energy on earth with a 5GW compute cluster; and sovereign capital — MGX, Mubadala, G42 — is already building non-dollar rails for AI. Nowhere else co-locates the regulator, the energy, the compute and the capital in one jurisdiction. So I'm relocating to build it on the ground, from day one.

I don't need a thousand participants on day one. I need two great ones — and a market they can both trust.

That's the whole plan, honestly. Start with one energy or datacenter partner and one AI startup. Broker the first deals by hand. Standardize the Compute-SAFE through the first ten. Then open the marketplace and the secondary. A ten-year build, done in the open, examined at every step. If you own compute or energy, if you're a founder tired of the cash-then-spend loop, or if you invest in the AI economy and want it denominated in more than dollars — I'd like to talk.