OpenAI, Anthropic AI Lead Is ‘Very Perishable,’ Professor Says - Alphabet (NASDAQ:GOOGL)
OpenAI and Anthropic may remain ahead of Chinese rivals in artificial intelligence and still find that technological lead worth less than expected, according to Howard Yu, LEGO Professor of Management and Innovation at IMD Business School.
Asked about a roughly four-to-six-month lead for U.S. frontier labs, Yu called such an advantage "very perishable."
"Most of the use case, you don’t need that," Yu told Benzinga.
A Six-Month Lead May Not Matter
Yu’s argument is not simply that Chinese labs could catch up. It is that many customers may not need the extra capability in the first place.
He pointed to a demonstration at Tencent Holdings Ltd. (OTC:TCEHY), where a WeChat agent scanned medicine packaging and quickly produced personalized guidance using information already held inside the platform.
"You don’t need extreme AI model to deliver useful information," Yu said.
The same principle applies to businesses, he said, where many AI workloads remain relatively simple.
"Most of the company using AI is for simple stuff," Yu said. "You don’t need to drive a Ferrari to go to Walmart."
Cost matters beyond choosing between models. Yu said executives he works with also weigh spending on AI tokens against simply hiring more people.
"Should we buy more token or should I hire more human hands?" he said.
Citing research from MIT and McKinsey, Yu said many corporate AI deployments have also struggled to deliver the expected returns.
Chinese Models Add More Pressure
Cheaper Chinese open-source models create a separate challenge for frontier labs, even if OpenAI and Anthropic remain technically ahead.
Yu said businesses can download those models, run them inside their own data centers and retain greater control over company information. He said demand for independence from the U.S. technology stack is particularly strong in Europe but is also appearing among U.S. companies.
Yu said enterprises are already using Chinese open models, including a well-known Andreessen Horowitz-backed Silicon Valley startup.
Intel Corp. (NASDAQ:INTC) CEO Lip-Bu Tan made a similar point this week, saying models including DeepSeek, Kimi and GLM were already "good enough," with lower token costs and "very fast" adoption.
Consumers May Be Different
Yu does not expect that pressure to dislodge OpenAI or Anthropic as easily in the Western consumer market.
Their stickiness, he said, may have less to do with having the smartest model than with simple product advantages such as easy access and a clean interface.
Enterprise customers are different. Yu described B2B as a much larger and more sophisticated market, where technical users can deploy open models without needing the same polished consumer interface.
U.S. Labs Still Favored to Lead
Prediction-market traders still expect U.S. companies to own the frontier. On Polymarket, Anthropic and Alphabet Inc. (NASDAQ:GOOGL) are nearly tied to have the best AI model at the end of 2026, at about 43% and 42%, respectively.
OpenAI sits near 8% and Meta around 4%, while Chinese labs remain around 1% or lower. About $2.5 million has been traded.
That market is asking who builds the most capable model. Yu’s argument is that businesses face a different question: whether the extra performance is valuable enough to pay for.
Image: Shutterstock
If you had to go and find out, you found out late. Benzinga Pro files the headline, speaks it aloud and flags the ticker as it happens. The whole product is the difference between learning something and being told.
Try it free for 14 days →© 2026 Benzinga.com. Benzinga does not provide investment advice. All rights reserved.
To add Benzinga News as your preferred source on Google, click here.