📰 Key Takeaways

At Baidu World 2026 in Beijing on November 12, Baidu founder Robin Li said commercial AI adoption among enterprises is “only just starting.” He urged companies to adopt AI agents more broadly, even as concerns keep growing over ballooning token budgets and the technology’s potential to hurt jobs and push up unemployment.

The report says companies in sectors like finance and manufacturing are rethinking their existing workflows as AI spending climbs. The event also showed off Baidu’s “Tianchi 1024-card supernode” AI computing system, the hardware infrastructure behind large-scale model compute.

That said, the original summary is very short. It only covers Li’s core message and the event backdrop. It doesn’t give specific spending figures, token usage numbers, real-world agent use cases, or quantified data on job impact, so we can’t dig further into the mechanics or details. See the original link for the full story.

💬 JudyAI Lab Take

At Baidu World 2026, Baidu founder Robin Li said commercial AI adoption among enterprises is “only just starting,” and he called on companies to adopt AI agents more widely. What’s worth noting is that he made this case against a backdrop of ballooning token budgets and rising worries about jobs.

This story points to a tension in the AI industry: the supply side is pushing hard, while the demand side is still doing the math. The report says companies in sectors like finance and manufacturing are rethinking existing workflows as AI spending climbs. That suggests adopting AI is no longer just about adding another tool. Companies are redesigning the workflows themselves. Meanwhile, the “Tianchi 1024-card supernode” shown at the event signals that large-scale compute infrastructure is still where vendors compete. For AI builders, this means whether an agent actually lands depends not only on model capability, but also on whether costs are controllable and whether a workflow is worth rewriting. When budgets and job worries are both under the microscope, products that can clearly explain their input-versus-output will be an easier sell to enterprises than ones that only talk about technical potential. The original piece doesn’t give specific spending or usage numbers, so you’ll need to check the full report for details.

Here’s a small exercise: for the AI app or agent workflow you’re building, list the token cost per run and the manual steps it replaces, so you can see clearly how the math works out.

📅 Source Info

🔗 Further Reading