DeepMind CEO Dismisses DeepSeek $6M AI Claim – Bloomberg
DeepMind CEO, Demis Hassabis, has dismissed claims made by Chinese AI startup DeepSeek regarding the cost of developing its artificial intelligence system, calling the reported figure of $5.6 million “exaggerated and a little bit misleading.”
This follows DeepSeek announcement last month that it had released a popular chatbot and AI model for far less than the costs reportedly spent by U.S. companies like DeepMind and OpenAI.
At the Artificial Intelligence Action Summit in Paris, Hassabis told Bloomberg Television, “They seem to have only reported the cost of the final training round, which is a fraction of the total cost.” He explained that the $5.6 million likely only covers the last stage of the AI model’s development, which represents a small portion of the total expenditure.
Bloomberg News also reported that OpenAI and Microsoft are investigating whether a group associated with DeepSeek used a method known as “distillation” to acquire data from OpenAI models. Distillation involves one model extracting outputs from another for the purpose of training.
Hassabis suggested that DeepSeek may have used Western models for this purpose, but did not provide further details on the matter. He added, “They seem to have only reported the cost of the final training round, which is a fraction of the total cost.”
READ ALSO: Chinese AI Startup DeepSeek Disrupts Global Tech Industry
DeepSeek’s claim that it used older Nvidia Corp. chips for its AI model has also raised questions among researchers and industry experts. Reports indicate that U.S. authorities are looking into whether DeepSeek bypassed restrictions on chip sales by purchasing semiconductors via Singapore.
Hassabis dismissed the notion that DeepSeek approach has disrupted the economics of AI development. “We don’t see any new silver bullet technologies. DeepSeek is not an outlier on the efficiency curve,” he said. He also noted that Google’s own AI model, Gemini, is more efficient in terms of training-to-performance and cost-to-performance ratios.
Microsoft has also committed to investing approximately $80 billion in AI-enabled data centres to train models worldwide, with Vice Chairman Brad Smith detailing the company’s strategy to counter China’s growing presence in the tech industry.
Alphabet, Google’s parent company, is also continuing to invest heavily in AI. Last week, the company announced plans to spend $75 billion on capital expenditures in 2025, focusing on its cloud-computing division and AI services, including its Gemini model, which is being integrated into Google’s search and other products.
