DeepSeek
Chinese lab that shocked the market with cheap frontier models
DeepSeek, funded by quant fund High-Flyer, stunned the industry in early 2025 with DeepSeek-V3 and the R1 reasoning model, achieving near-frontier performance at a fraction of the training cost and triggering a sharp Nvidia stock selloff. It trains under US export-control constraints (restricted to H800/H20-class chips) and releases open weights. It is the leading demonstration that algorithmic efficiency can offset compute restrictions.
Flagship models
DeepSeek-V3, R1
Backer
High-Flyer (quant fund)
Constraint
Export-controlled chips
Impact
Jan 2025 Nvidia selloff
How it fits the stack
DeepSeek with what it depends on (above) and what it feeds (below). The figure renders as a crawlable diagram and upgrades to an interactive 3D graph as it scrolls into view.
DeepSeek in the AI stack. DeepSeek with its immediate upstream dependencies (top) and downstream dependents (bottom) in the AI value chain. Hover a node in 3D, or read the full relationships below.
Graph data (text) — 10 entities, 9 relationships
- DeepSeek —depends on→ China AI Chip Market / H20 & Domestic Substitution
- DeepSeek —depends on→ GPU gray market / diversion
- DeepSeek —uses→ Huawei Ascend (910B/910C)
- DeepSeek —depends on→ Nvidia Data-Center GPU (Blackwell/Rubin)
- DeepSeek —uses→ Synthetic data & distillation
- DeepSeek —competes with→ Alibaba Qwen (Tongyi)
- DeepSeek —competes with→ OpenAI
- DeepSeek —competes with→ StepFun (Jieyue Xingchen)
- DeepSeek —designs→ DeepSeek R2 / V3.x
Depends on ↑ · 5
Context — capital, rivals, policy · · 6