Jefferies Warns of AI Capital Destruction Amid Chinese Competition

The brokerage firm Jefferies warns that the United States faces a risk of massive capital destruction as cheaper, open-source Chinese AI models erode the market dominance of US hyperscalers. Investors are increasingly questioning the returns on hundreds of billions of dollars invested in AI infrastructure. The four largest US hyperscalers are projected to spend $695 billion in capital expenditure in 2026 and $870 billion in 2027.

This financial risk is heightened by a collapse in AI token pricing and the sector's position as the largest issuer of investment-grade debt in the US. Jefferies argues that China has become a technological peer to the US, citing the launch of Moonshot AI's Kimi K3 model. Data from OpenRouter shows Chinese models processed 36.39 trillion tokens in the week ending July 19, significantly outperforming the 7.39 trillion tokens processed by leading US models.

Among the hyperscalers, Alphabet Inc. has already increased its 2026 capital expenditure guidance by $15 billion, bringing its projected range to $195-205 billion. Investors are currently awaiting updated expenditure guidance from Microsoft, Amazon.com, and Meta Platforms Incorporated.