US firms cut flagship AI spend for cheaper models including Chinese ones
The Wall Street Journal reports that major US corporations are dialing back reliance on expensive OpenAI and Anthropic flagship models for routine work. Companies are moving to an a la carte mix that includes lower-cost alternatives, among them Chinese-built models, to balance performance and spend.
Key takeaway
Buyer behavior is bifurcating: premium US models stay reserved for hard reasoning while commodity workloads migrate to cheaper, often Chinese, options.
Context
According to the WSJ account summarized in the candidate, corporate buyers that previously defaulted to top-tier US assistants are now scrutinizing unit economics on everyday tasks. That push is reducing automatic spend on the most expensive OpenAI and Anthropic offerings when cheaper models are judged good enough.
The reported pattern is not a single contract cancellation but a procurement shift toward mixing providers by task. If it sticks, it pressures premium labs to prove where higher prices still buy material quality, and it raises the strategic profile of lower-cost and Chinese open-weight alternatives inside US enterprises.