Company profile · 4 min read
Liquid Compute is building market infrastructure for transferable compute capacity
Liquid Compute is developing a physical compute market with verification, settlement, reference indices and planned derivatives for infrastructure participants.
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
Published · Updated
Liquid Compute is for the companies that need to price, transfer or finance compute—not for a startup looking for another ordinary GPU rental marketplace. It is building market infrastructure around verification, settlement, reference prices and eventually derivatives.
What it does
The current site describes a physical order book for compute, with paths to buy or sell capacity, trade GPU indices and finance compute. It says the market is intended for neoclouds, AI companies and trading firms, and that access is being onboarded. Liquid Compute homepage
The company’s reasoning is concrete: GPU capacity is heterogeneous, location- and workload-dependent, and unused capacity cannot be stored for later. Liquid Compute wants standard contracts and a common framework so capacity can be transferred without every resale requiring a new contract, credit decision, payment setup and SLA. Liquid Compute launch
The important status distinction
Liquid Compute is not presenting a fully approved public derivatives venue. The homepage says PMEX Markets is pending designation as a Designated Contract Market and registration as a Derivatives Clearing Organization by the CFTC. The YC launch page likewise describes pending applications. That is a central product fact: regulatory approval, cleared trading and market liquidity should not be inferred from the “regulated exchange” positioning.
The launch announcement says the company emerged from stealth with a $15 million seed round co-led by Chemistry and FirstMark. It does not publish a customer price list or a standard transaction schedule. Liquid Compute announcement The YC profile identifies Ronit Jain as CEO and Aarav Patel as CTO, with the founders working on the exchange.
Who should look
The strongest fit is a neocloud, data center, broker or trading firm that already has a measurable exposure to compute utilization, delivery or financing. The tradeoff is that a new market needs standardized definitions, reliable supply, counterparties and liquidity at the same time. Before treating it as hedging infrastructure, ask what the contract specifies, how performance is verified, who clears it, what happens on delivery failure and what the regulatory timeline permits today.
My editorial take
I would monitor Liquid Compute as a potentially important layer for the AI infrastructure economy, but I would not treat the current site as evidence that a founder can immediately hedge compute exposure there. For a buyer, the next step is a market-access conversation focused on contract terms and current counterparties. For a routine AI workload, buying capacity directly remains the simpler decision until the exchange, approvals and liquidity are demonstrably in place.
Quick facts
| Field | Sourced detail |
|---|---|
| Buyer | Neoclouds, data centers, AI companies, brokers and trading firms |
| Product | Physical compute market with reference indices and planned derivatives |
| Regulatory status | DCM designation and DCO registration described as pending |
| Funding signal | Company announced a $15M seed round |
| Main fit question | Do you need market infrastructure, or simply compute capacity? |
Sources checked
Liquid Compute’s YC profile, homepage, launch announcement and about page were checked on 2026-09-19.
Cohort context
Liquid Compute is listed in Winter 2025. In our 2026-09-18 directory snapshot, 12 of 165 listed companies in that cohort have YC’s primary industry label Fintech (7.3%). This is a current-directory comparison, not an original intake count or a performance ranking. Nine-cohort dataset.
Public website snapshot
Observed 2026-09-19T16:19:37.905Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.
| Signal | Homepage observation |
|---|---|
| Product description metadata | Observed |
| Canonical link | Observed |
| H1 or H2 heading | Observed |
| Typed structured data | Not observed in this response |
| Docs/developer link | Not observed in this response |
| Pricing link | Not observed in this response |
| llms.txt link | Not observed in this response |
| Markdown alternate | Not observed in this response |
Public observations · Collection method. Missing links here do not establish that a capability or file is absent elsewhere.
About the author
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.
First1000 ↗ · X ↗