SF Compute's Evan Conrad: AI startups hit credit limits before power limits
AI Engineer · youtube · 2026-10-23
AI Engineer interview with San Francisco Compute co-founder Evan Conrad on how a struggling AI lab that subleased its own GPU cluster to survive became both a cluster operator and a compute marketplace.
Key arguments:
- Problems facing new labs and inference companies—sold-out clusters, deposits draining bank balances, late and broken hardware, unpredictable demand—are all one risk problem
- AI companies hit credit limits before power limits: deposits scale from $10M toward $1B, keeping startups out
- Without a working compute market, the price of entry keeps climbing until no new startups can get in
- He argues for physically settled compute markets and explains why his firm must operate clusters itself, and why nobody else does physical settlement
Bottom line: the missing compute market is choking the AI startup ecosystem; market-based risk sharing is what keeps new entrants alive.
More from Venture
- Agentic data toll: enterprise agent data services spend to hit ~$30B by 2030, Bajarin estimates — BenBajarin · 2026-10-23
- Personal AI agents become a capital-intensive race as cloud execution environments reshape cost structures — sujingshen · 2026-10-23
- DeepSeek V4.1 Flash matches Gemini 3.1 Pro on ARC-AGI-2 at ~1/6 the cost — teortaxesTex · 2026-10-23
- Arbitrage idea: take faceless YouTube gigs on Fiverr, outsource via Xianyu with an agent — yangyi · 2026-10-23
- Stripe's Link adds purchase protection as AI agent "fat-finger" insurance debate heats up — jeff_weinstein · 2026-10-23
- VC Yang Yi: don't build AI apps now, the frontier period belongs to infra plays — yangyi · 2026-10-23