November 17, 2026, (Inside AI) — Baidu CEO Robin Li cautioned on Tuesday that the artificial intelligence industry may be inflating a bubble if companies cannot convert adoption into higher profits. Speaking at the APEC CEO Summit in Shenzhen, Li said the spread of AI applications has fallen short of expectations despite heavy spending on infrastructure, rapid model development and rising token use.
His warning lands at a delicate moment. AI investment has surged worldwide, yet measurable returns remain elusive for many enterprises. Li's remarks suggest the gap between spending and profit is widening, not closing.
Li did not single out any company or region. But his venue choice mattered. The APEC summit draws business and government leaders from across the Asia-Pacific, a region that has become a central battleground for AI infrastructure. Shenzhen itself hosts some of China's most aggressive AI buildouts.
The core problem, as Li framed it, is organizational. Companies need to rethink how they structure work to turn AI capabilities into business gains. He pointed to changes Baidu has made internally, building on a theme he raised at Baidu World last week about how companies can organize around AI.
Read: Jensen Huang's Exit Exposes $105 Billion Nvidia Guarantee Risk
That internal focus is notable. Baidu has invested heavily in AI models and cloud infrastructure. If even a company with its own large language models is still searching for profit, the challenge is likely structural rather than technical.
Token use has risen sharply across the industry. That metric often signals growing engagement with AI models. But Li's comments imply that higher token consumption does not automatically translate into higher revenue or productivity. The two can diverge.
Infrastructure spending has also accelerated. Data centers, chips and energy contracts have absorbed billions in capital. Yet application adoption has lagged. That mismatch is the classic signature of a bubble: investment runs ahead of real demand.
Li's warning echoes earlier caution from other industry figures. But it carries extra weight coming from the CEO of a company that operates across search, cloud and autonomous driving. Baidu has direct exposure to both the infrastructure and application sides of AI.
His proposed remedy is organizational, not technical. Companies must redesign workflows, roles and decision-making around AI. That is a slower and harder task than buying more compute. It also demands clearer metrics for success.
Read: Nasdaq Futures Dip as Nvidia CEO Exit Rattles AI Investors
For now, Li's message is a reminder that AI's commercial promise remains unproven at scale. The industry may need to prove it can generate profits before it can justify another round of spending.
Shamil has a low tolerance for AI that only works in controlled demos. He covers agentic systems and generative AI with the skepticism of someone who has seen too many “revolutionary” tools quietly disappear six months after launch.