DeepMind CEO says DeepSeek’s cost claims ‘exaggerated,’ Bloomberg reports

Google (GOOGL) DeepMind CEO Demis Hassabis says DeepSeek’s $5.6M training cost claim is “exaggerated and a little bit misleading,” Bloomberg’s Yazhou Sun and Tom Mackenzie report. Hassabis, who runs the AI unit of Google, told Bloomberg Television that DeepSeek “seems to have only reported the cost of the final training round, which is a fraction of the total cost.” DeepSeek is a Chinese-built large-language open-source model that claims to rival offerings from Microsoft-backed (MSFT) OpenAI’s ChatGPT and Meta Platforms (META) but using a much smaller budget. OpenAI and Microsoft Corp. are investigating whether a group tied to DeepSeek obtained data from OpenAI using a process known as distillation, and Hassabis told Bloomberg TV that DeepSeek seems “to have relied on some Western models to distill from,” without offering specifics.