Cheap Chinese and open source LLMs could trigger trillion-dollar AI infrastructure 'reset'
Jefferies wonders if LLMs will ever be profitable during the 'ongoing collapse in token pricing'
A financial analyst has raised concerns that the decline in AI costs, due to the emergence of low-cost Chinese and open-source models, could lead to a "reset" in the AI industry. Jefferies estimates that half of the $2.3 trillion contracted backlog at major cloud providers like Microsoft, Alphabet, Amazon, and Oracle is tied to OpenAI and Anthropic.
The rapid growth of Chinese models, now processing 76.12 trillion tokens in the past week compared to just 4.37 trillion last April, poses a threat to the pricing power of the US leaders. The launch of open-source Kimi K3 from Moonshot AI is particularly impressive, as it matches the performance of Anthropic's Claude Fable 5 at a relatively low cost.
While the hyperscalers are planning to collectively invest $600-$800 billion in new datacenters this year, the analyst questions where their future revenue streams will come from, given the continued decline in token pricing. The long-term outlook for AI seems promising, but the medium-term outlook is uncertain, according to Jefferies.
Written by urgent.news from The Register Science's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
