Emad Mostaque
Founder of Intelligent Internet · Founder of Stability AI (former) · Author of The Last Economy
aka Imad
“building great solid models is cutting edge manufacturing... why are Chinese EVs better than Ford's? This actually feels like the same thing... they're just executing that manufacturing process with what they have.”
“Kimik 3 is $15, Sonnet is 20, or Opus is $40... when you back out the numbers from the Chinese models and the chips they're running on, they're probably making 80, 90% margins now... we will see the cost of K3 drop by 10 to 50 times.”
“American companies like Modal, like Fireworks, like Base 10 will be able to serve this model 10 times cheaper than their Chinese competitors because they have access to the Nvidia and AMD big chips.”
“Why do Americans pay more for drugs than everyone else? All the R and D happens in America, you pay the premium, just like tokens, premiums. And then what happens? You have generics elsewhere.”
Emad Mostaque
One-line summary: Founder of Intelligent Internet and former founder of Stability AI; tracked for open-weight economics, inference-cost and quantization calls.
What they're known for
Founder of Stability AI (creator of Stable Diffusion, the open image model that seeded a 100M+-download ecosystem) and now founder of Intelligent Internet; author of The Last Economy. Appears as the emergency-pod expert guest on the Kimi K3 release, tracking the Chinese open-weight labs closely (WhatsApp back-channel with Diamandis).
Why they matter to artificial-intelligence
Mostaque is the thread's sharpest voice on open-weight economics and the mechanics of Chinese frontier catch-up. His load-bearing claims: (1) frontier model-building is now "cutting-edge manufacturing" (the Chinese-EV analogy) — data mix and engineering, not architectural magic; (2) Chinese open models price high (Kimi K3 at $15/M) because they run 80-90% margins on inferior Chinese silicon, and US inference providers (Modal, Fireworks, Base10) will serve the same open weights ~10x cheaper on Nvidia/AMD; (3) the drug-generic analogy for the closed-vs-open value split. He also carries the quantization/edge thread (Bonsai 27B ternary, Tencent binary, distillation-to-edge). See chinese-open-weight-frontier-parity, open-source-share-shift-bullish-for-compute, edge-inference-shift.
Said
Speaker-attributed claims extracted from diarized sources. Each bullet mirrors one entry in quotes: frontmatter — keep them in sync.
-
On chinese-open-weight-frontier-parity:
"building great solid models is cutting edge manufacturing... why are Chinese EVs better than Ford's? This actually feels like the same thing... they're just executing that manufacturing process with what they have." — 2026-07-19-podcast-moonshots-urgent-update-ai-sputnik-moment-kimi-k3-released (2026-07-19)
-
On open-source-share-shift-bullish-for-compute:
"Kimik 3 is $15, Sonnet is 20, or Opus is $40... when you back out the numbers from the Chinese models and the chips they're running on, they're probably making 80, 90% margins now... we will see the cost of K3 drop by 10 to 50 times." — 2026-07-19-podcast-moonshots-urgent-update-ai-sputnik-moment-kimi-k3-released (2026-07-19)
-
On cuda-moat-erosion-at-inference:
"American companies like Modal, like Fireworks, like Base 10 will be able to serve this model 10 times cheaper than their Chinese competitors because they have access to the Nvidia and AMD big chips." — 2026-07-19-podcast-moonshots-urgent-update-ai-sputnik-moment-kimi-k3-released (2026-07-19)
-
On open-vs-closed-source-model-economics:
"Why do Americans pay more for drugs than everyone else? All the R and D happens in America, you pay the premium, just like tokens, premiums. And then what happens? You have generics elsewhere." — 2026-07-19-podcast-moonshots-urgent-update-ai-sputnik-moment-kimi-k3-released (2026-07-19)