Jefferies strategist sees risk of heavy losses in US AI spending
Jefferies equity strategist Chris Wood warned that the US AI buildout could lead to large losses on invested capital as inexpensive Chinese open-source models gain share and AI output prices fall. He pointed to hyperscaler capital expenditure guidance of about $695 billion for 2026, rising to $870 billion in 2027, and said much of the spending is being financed with debt rather than cash. Wood also noted that Chinese AI models processed 36.39 trillion tokens in the week ending July 19, 2026, compared with 7.39 trillion for top US models.
In an Oct. 9, 2026 note, Jefferies equity strategist Chris Wood argued that America's AI infrastructure push may destroy capital rather than reward it. He cited hyperscaler capital spending guidance near $695 billion for 2026, climbing to about $870 billion in 2027, with debt increasingly replacing cash to fund it.
Wood also highlighted usage and pricing: Chinese models handled 36.39 trillion tokens in the week ending July 19, 2026, versus 7.39 trillion for leading US models. Cheaper options such as Kimi K3 and GLM-5.2, at roughly a quarter per token of some US rivals, could undercut returns. He favored exposure to chipmakers and other enabling suppliers.
If AI capital spending slows or losses mount, investors and lenders exposed to hyperscalers could face weaker returns and tighter credit. Tech workers and communities tied to data-center construction may see hiring or local spending cool. Businesses and consumers might benefit if cheaper Chinese open-source models lower AI costs, though reliance on foreign models could raise security and governance questions.