Cheap Chinese and open‑source LLMs may trigger trillion‑dollar AI infrastructure reset
amazon anthropic microsoft openai open-source
| Source: Mastodon | Original article
Cheap Chinese and open‑source LLMs could force a trillion‑dollar AI infrastructure reset, as Jefferies notes half of the $2.3 trillion backlog of major cloud providers is tied to OpenAI and Anthropic.
A Jefferies research note released this week warns that a wave of low‑cost Chinese large‑language models (LLMs) and open‑source alternatives could force a “reset” of the AI infrastructure market, jeopardising up to a trillion dollars of cloud‑provider capital spending.
The analyst team, focusing on Asian markets, estimates that the four biggest cloud operators – Microsoft, Alphabet, Amazon and Oracle – carry roughly $2.3 trillion of contracted AI capacity. About half of that backlog is tied to OpenAI and Anthropic services. Jefferies argues that the arrival of Chinese models that cost 10‑60 times less than their Western counterparts, together with open‑weight offerings that undercut prices by 60‑90 percent, could erode the revenue streams that these providers rely on to recoup the massive capex behind their AI platforms.
If the pricing pressure materialises, cloud firms may struggle to meet the financial expectations built into their multi‑year AI contracts, potentially prompting a re‑evaluation of spending plans, pricing structures and even the valuation of frontier‑AI IPOs that hinge on continued growth. The note stresses that the forecast is based on the analysts’ own calculations rather than an imminent event.
Stakeholders will be watching how quickly the cheaper models gain market share, whether major cloud players adjust pricing or accelerate the rollout of their own cost‑efficient architectures, and how investors react to the prospect of a $2‑trillion‑scale shift in AI spend. The next few quarters should reveal whether the “reset” remains a theoretical risk or becomes a defining moment for the AI infrastructure ecosystem.
Sources
Back to AIPULSEN