Generated by Codex with GPT-5

This summary covers The Economist’s May 9th, 2026 Finance & economics article on artificial intelligence and China’s office-property market, published under the headline The rent is too damn AI and listed in the contents as AI rescues Chinese landlords.

The article uses a neat irony to explain a real-estate rebound. A court in Hangzhou, one of China’s leading AI hubs, ruled that firms cannot simply dismiss workers and replace them with artificial intelligence. That is good news for employees. It is also good news for commercial landlords, because office towers do not earn rent from software agents. They need people in seats, and China’s newest source of people in seats is the AI boom.

The point is not that AI has solved China’s property crisis. It has not. The article’s sharper claim is narrower: in a few technology-heavy cities, demand from AI firms is helping absorb some of the empty office space left behind by the collapse of the old real-estate cycle and the earlier crackdown on internet companies.

Empty Towers, New Tenants

China’s property bust is usually discussed through housing, where vast numbers of flats sit empty. The article shifts attention to commercial property. Years of overbuilding left many cities with too much office space, and projects begun before the pandemic have continued to come onto the market even as demand weakened. In some inland cities, more than 40% of top-grade offices are vacant.

Hangzhou and Shenzhen show how the problem developed in the most advanced parts of the economy. Both cities were hit when China’s tech crackdown and the property downturn arrived around the same time. Large employers such as Alibaba and Tencent cut jobs, and the loss of technology tenants pushed up vacancies in premium offices. In Hangzhou and Shenzhen, high-end office vacancies rose from roughly one-fifth in 2019 to about three-tenths at their worst.

AI has changed that trajectory, at least for now. Demand from AI firms across China was three times higher in March than a year earlier, according to UBS. That increase started from a low base, but it matters because AI activity is clustered in a small number of cities. Hangzhou and Shenzhen are precisely the sort of places that benefit. Their grade-A office vacancy rates have fallen back toward about 20%.

DeepSeek’s rise is part of the story, but not the whole story. The article notes that some insiders joked that the lab might revive Hangzhou’s real-estate market by itself. In practice, the rebound has broader support. Alibaba is trying to recast itself as an AI company. Local officials are subsidizing rents for innovative firms. New developments such as Westlake 66 expect technology companies to occupy much of their office space. The AI expansion also pulls in accountants, consultants, lawyers and other professional-service firms that follow growing companies.

Policy is helping on the supply side, too. Local authorities have started limiting new commercial real-estate supply and selling less land for development. If demand continues to improve while supply tightens, landlords may finally regain some pricing power.

The Catch

The article’s closing warning is that the same force rescuing landlords may later undercut them. Much of the new demand is not just for software engineers. AI firms also need support staff and back-office workers, at least during their current expansion. Those are exactly the kinds of roles that AI agents are expected to automate over time.

That creates a fragile kind of recovery. In the short run, an AI boom means more companies, more hiring and more office demand in the cities where the industry clusters. A court ruling that slows direct replacement of workers with AI may even extend the effect. In the longer run, however, AI could reduce the number of clerical, administrative and quality-checking jobs that make office floors valuable in the first place.

The article therefore presents China’s AI office boom as a reprieve rather than a cure. It shows how a new technology cycle can soften the damage from an old property cycle, especially in cities with the right industrial base. But it also shows why landlords should be careful about celebrating too early. If AI’s next phase creates companies that need fewer employees per unit of revenue, the desks now being filled could empty again.

The broader takeaway is that China’s property market is no longer moving as one giant story. Residential oversupply, weak smaller cities and distressed developers remain serious problems. Yet in a few AI centers, a different pattern is emerging: technology firms are giving premium offices a second chance. Whether that chance becomes a durable recovery depends on whether AI remains a job creator for these cities or becomes the tool that lets their tenants rent less space.