DeepSeek trained models on banned Nvidia Blackwell chips
DeepSeek has apparently circumvented US export controls to train advanced AI models on Nvidia's Blackwell processors, with Google, OpenAI, and Anthropic actively preparing for a major competitive challenge.
DeepSeek has apparently circumvented US export controls to train advanced AI models on Nvidia's Blackwell processors, which are prohibited from sale to China. The disclosure comes as Google, OpenAI, and Anthropic are actively preparing for a major DeepSeek release, signaling the Chinese AI company is preparing a competitive challenge to US-dominant AI labs. The timing and scope suggest another market-shaping announcement may be imminent.
This represents a direct challenge to US semiconductor restrictions designed to maintain American technological leadership in AI. DeepSeek's ability to access banned hardware suggests either sophisticated workarounds to export controls or gaps in enforcement. The response from Tier 1 US labs—proactively bracing for DeepSeek's release rather than responding reactively—indicates the threat is taken seriously at the highest levels of the industry.
Sources
- T2
- T2
Stay informed. The best AI coverage, delivered weekly.