China AI speaker says 2026 compute spend may stay under RMB 20B
teortaxesTex · x · 2026-07-23
A quoted talk says China will not try to compete with the US at the largest model scale yet, and instead focus first on model sizes it can afford to train and run.
Key points from the excerpt:
- The speaker expects no more than 20B RMB spent on compute in 2026.
- He says the largest current model activates about 800B parameters.
- Training that scale would require roughly 50K GB300s or 200K Huawei 950s.
- The quote argues you can brute-force a model that large, but not do enough research at that scale beforehand.
- He also says the remaining gaps in foundation-model competition are mostly cost, time, and user experience.
Related event: DeepSeek Founder's Investor Call Reveals AGI Roadmap and Compute Plans(37 posts)→
More from Infra
- Engram's random reads don't suit SSDs; CPU-memory over NVLink could serve all 72 GPUs — bookwormengr · 2026-09-11
- 80% of the DIY LLM inference hype posters have already quit — it's brutally hard systems work — abhijithneil · 2026-09-11
- Hugging Face's Ultra Scale Playbook: a free book on training LLMs on GPU clusters — mdancho84 · 2026-09-11
- Is inference latency becoming the biggest bottleneck for production AI agents? — Euphoric_Sea632 · 2026-09-11
- LLM Serving Metrics Thread: Why TPOT and Uptime Make or Break User Experience — abhijithneil · 2026-09-11
- PlanetScale launches sharded Postgres: 768 servers acting as one, 1PB scale — dhruv2038 · 2026-09-11