source · blog
On DeepSeek and Export Controls
src_amodei_deepseek_export_controls
https://darioamodei.com/post/on-deepseek-and-export-controls
authors: Dario Amodei
published: 2025-01
accessed: 2026-04-19
Notes
Blog prior is 0.35; bumped because author is Anthropic CEO with direct knowledge of frontier training costs and chip supply. Still discounted for advocacy framing (explicit policy argument for export controls).
Intake provenance
- method
- httpx
- tool
- afls-ingest/0.0.1
- git sha
- 604c9dfd252a
- at
- 2026-04-19T18:48:45.102779Z
- sha256
- 9aa01480a927…
Evidence from this source (6)
- weight0.75
method: expert_estimate · locator: DeepSeek's Models section
“DeepSeek-V3 is not a unique breakthrough or something that fundamentally changes the economics of LLM's; it's an expected point on an ongoing cost reduction curve.”
- weight0.60
method: journalistic_report · locator: DeepSeek's Models section, SemiAnalysis reference
“It's been reported --- we can't be certain it is true --- that DeepSeek actually had 50,000 Hopper generation chips, which I'd guess is within a factor ~2-3x of what the major US AI companies have... Those 50,000 Hopper chips cost on the order of ~$1B.”
- weight0.80
method: expert_estimate · locator: DeepSeek's Models section, paragraph on cost curve
“DeepSeek produced a model close to the performance of US models 7-10 months older, for a good deal less cost (but not anywhere near the ratios people have suggested).”
- weight0.70
method: expert_estimate · locator: Export Controls section, opening bullets
“Making AI that is smarter than almost all humans at almost all things will require millions of chips, tens of billions of dollars (at least), and is most likely to happen in 2026-2027.”
- weight0.75
method: expert_estimate · locator: Three Dynamics section, #3 Shifting the paradigm
“Spending $1M instead of $0.1M is enough to get huge gains... we're at a unique 'crossover point' where there is a powerful new paradigm that is early on the scaling curve.”
- weight0.65
method: expert_estimate · locator: Footnote 9, quoting Amodei/Pottinger op-ed
“China's best AI chips, the Huawei Ascend series, are substantially less capable than the leading chip made by U.S.-based Nvidia... There's not a single noteworthy cluster of Huawei Ascend chips outside China today.”