Washington University CIO Warns AI Labs May Struggle Against Cheaper Chinese Models

Scott Wilson, chief investment officer for Washington University's endowment, said the large-scale AI companies OpenAI and Anthropic may find it difficult to sustain their high spending as Chinese firms release cheaper, open-weight models with comparable performance. Wilson, who previously profited from an early investment in SpaceX, voiced his concerns at a Fort Worth gathering of asset managers and tech founders, describing the AI sector as a "wild West" of rapid spending and inflated valuations.

He noted that Chinese providers such as DeepSeek, Alibaba's Qwen, Zhipu AI and Tencent have narrowed the performance gap while offering significant cost advantages. Data from the developer platform OpenRouter shows DeepSeek handling 25.3% of text-model requests, surpassing OpenAI's 18.6% and Anthropic's 2.9%.

"These trillion-dollar-plus frontier companies are not worth the liabilities that they signed up for," Wilson said. "There’s going to be a ton of free alternatives."

Wilson added that portfolio companies he works with are increasingly adopting open-source solutions to reduce expenses, likening the shift to competition from low-cost imports.

Venture investor Vinod Khosla, an early backer of OpenAI, disagreed, arguing that closed-source firms can lower overall costs by designing their own hardware and infrastructure.

"I'm not talking price, I'm talking about cost. From power to data center to chips, to infrastructure software to inference models, the cost of the stack is almost certainly going to be lower in closed-source models than open-source," Khosla said.

The contrasting views highlight an ongoing debate over whether the high-cost model-centric approach can remain viable if customers switch to more affordable alternatives.