New Blueprint For AI Labs
How should AI labs survive when API prices are deflating 10x a year? 📉 These three winning blueprints offer radical new paths forward.
What the video says
600 essays submitted for a high-stakes AI contest have yielded 3 winning blueprints for a next-generation world. Dwarkesh Patel announced the winners of his competition tackling the biggest questions in AI policy.
First-place winner Joshi Panu argues the OpenAI Foundation should dedicate tens of billions of dollars to end airborne pathogens. But a massive economic tension looms.
AI API pricing is deflating roughly 10 times per year due to open-source alternatives. This makes APIs structurally unprofitable for funding frontier training.
The third winning essay suggests labs adopt Hong Kong's 45-year-old rail plus property business model. Instead of relying on unprofitable APIs, labs must capture value from the surrounding ecosystem to survive this epoch-making transition.
