SF Compute CEO: OpenAI and Anthropic won't eat the world — every company will become a frontier lab
danshipper · x · 2026-08-26
Dan Shipper endorses SF Compute CEO Evan Conrad's contrarian thesis: OpenAI and Anthropic won't eat the world — everyone will look like OpenAI and Anthropic.
Conrad's arguments:
- Every company of meaningful scale will end up looking like a frontier lab; that's simply how companies work now.
- The belief that large labs have won with a compounding advantage assumes nobody else will do anything — a flawed default.
- The old bottleneck — only 100 people knowing how to train models — is dissolving.
Shipper adds that all of work is starting to look like AI research.
More from AGI Musings
- Terence Tao: Autoformalization still costly but rapidly improving — littmath · 2026-08-27
- Terence Tao: Use AI to produce better work, not just more PDFs — littmath · 2026-08-27
- Teaching students to keep their judgment when working with AI: the real frontier of AI education — DrKavner · 2026-08-27
- Phantom traffic jam experiment offers a grim model for AI progress — robleclerc · 2026-08-27
- Who can overrule your AI when it's confidently wrong? — YvesMulkers · 2026-08-27
- Brands are getting impressions in AI answers without the click — nikvassev · 2026-08-27