Analysis of DeepSeek's Compute Costs and Revenue Model
teortaxesTex · x · 2026-07-15
The author debunks conspiracy theories about DeepSeek (such as Singapore compute and subsidies) with math: at 10K tokens/s/GPU and $0.28/1M tokens, the annual revenue per GPU is about $88.3K. To hit $500M ARR, it only requires around 5,662 GPUs. This implies DeepSeek's business model is viable without needing conspiracy theories to explain it.
Related event: DeepSeek inference economics: sizing the GPU fleet behind its ARR(8 posts)→
More from Infra
- Ratel says it made agents 7x cheaper by loading only the tools each task needs — tensorqt · 2026-07-22
- Weaviate adds per-query profiling to pinpoint where a slow search query spends time — CShorten30 · 2026-07-22
- Mistral expands its Microsoft partnership as it adds more AI compute in Europe — MistralAI · 2026-07-22
- Sol-Engine Boosts Video Generation Speed by up to 5x with Training-Free Sparse Attention — songhan_mit · 2026-07-22
- PyTorch CTO to Explore Open Source AI Inference Economics and Workflow Optimization — PyTorch · 2026-07-22
- Tinkerers run GLM-5.2 at near-lossless quality on a $15,000 budget — amplifiedamp · 2026-07-21