Emad Mostaque: US firms can serve Kimi K3 at 1/10 the cost thanks to Nvidia, AMD chips
rohanpaul_ai · x · 2026-09-21
Stability AI co-founder Emad Mostaque, speaking on Peter Diamandis' YouTube channel, argues that US inference providers like Modal, Fireworks, and Baseten will be able to serve Kimi K3 at one-tenth the cost of their Chinese competitors because of access to advanced Nvidia and AMD chips.
He calls the situation ironic: much of the model's R&D has shifted to China, but once the model is optimized for next-generation hardware like Nvidia's Rubin architecture, its operating cost could fall by another 10 to 100 times.
More from Infra
- Data center debt built for Jane Street sours in secondary trading, yields hit ~11.3% — GaryMarcus · 2026-09-21
- China's CXMT starts mass-producing 11.95nm DRAM, 50% more dies per wafer — mark_k · 2026-09-21
- Huawei name-drops DeepSeek in keynote, but its actual compute share looks like a tiny fraction of Tencent's — teortaxesTex · 2026-09-21
- Custom CUDA shim runs Stable Diffusion on Mac faster than RTX 5090 on Windows, up to 61% quicker — LioDavinchy · 2026-09-21
- OpenAI preps new Codex Cloud built on Tailscale and Azure workload identity — ryanmerket · 2026-09-21
- Apple's C2 modem is competitive, and a cheaper Qualcomm deal may be next — Scobleizer · 2026-09-21