Did In depthThe AI model really costs a little less than $ 6 million to develop?
Demis HassabisGoogle’s artificial intelligence chief (AI), says he’s not so sure, Bloomberg News reported Monday (February 10).
In an interview with Bloomberg Television, Hassabis, who directs Google Depthcalled the idea that the Chinese startup spent so little to develop an AI system that competes with the “exaggerated and a little misleading” American technology giants.
Deepseek rocked the world of technology last month when he unveiled a chatbot and AI model that said he had developed expenses much less than companies like Openai. The launch caused a fall in many technological actions, destroying nearly 1 billion of dollars of market value.
Gokul naiduconsultant for SAPtold Pymnts last month that Deepseek “calls into question the story that innovation should have an unbearable cost”.
“For companies, this means that AI could soon be accessible to small and medium -sized businesses, not just technological giants with deep pockets,” added Naidu.
But Hassabis, interviewed at the top of the action of artificial intelligence in Paris, argued that Deepseek “seems to have only pointed out the cost of the final training series, which is a fraction of the total cost”.
He also rejected the idea that the rise of Deepseek shaken the economy of the development of artificial intelligence.
“We don’t see new Silver Bulte technologies,” said Hassabis. “Deepseek is not an aberrant value on the efficiency curve.”
His comments arise at a time when, as Pymnts wrote last week, the “arms race of AI becomes price costs AI.”
Meta’s budget for capital expenses could reach $ 65 billion, while Google has put aside $ 75 billion, mainly for data centers, servers and networking infrastructure. Amazon projects that it will spend $ 100 billion, while Microsoft reserves $ 80 billion to build data centers, train AI models and launch AI and Cloud applications.
Microsoft’s president Brad Smith wrote on the company’s blog last month that “artificial intelligence is the electricity of our time”, creating the next industrial revolution.
“However, AI requires heavy investments,” wrote Pymnts.
“The formation of large languages models uses thousands of GPUs (each NVIDIA GPU costs around $ 10,000 or more) or specialized AI chips for a total of tens or hundreds of millions of dollars. The execution of these large -scale AI models also requires high -performance data centers, which need more servers and require more cooling and maintenance. »»
