The AI investment narrative is shifting from model development toward infrastructure buildout. In its latest research note, UBS pointed out that power supply, data center capacity and financing capability are becoming the core variables driving the next stage of AI trading, while market attention has also moved from model competition to whether infrastructure can be delivered on time.

UBS energy and resources expert Catherine Gordon, in a recently published report, summarized this trend as "AI infrastructure constraints emerging as the next stage of the trade." She noted that the long-term power purchase agreements (PPAs) recently signed by Alphabet and Amazon further highlight the importance of power supply to AI expansion and keep infrastructure-type assets in the market spotlight.

Although the market still has concerns about the sustainability of AI capital expenditure and infrastructure financing, industry feedback continues to show that demand remains resilient. The key question now is not only whether AI demand can be realized, but also whether the supply side can keep pace with the speed of deployment.

Supply bottlenecks and execution capability will determine the next stage of AI investment

Gordon noted that although brief doubts over OpenAI's revenue expectations once triggered market volatility, the core of the AI investment logic still lies in whether infrastructure buildout can match demand. Power availability has drawn particular attention, and the long-term power purchase agreements signed by Alphabet and Amazon have further reinforced the market's focus on energy supply constraints.

Industry exchanges at the recent Data Center World Expo in Singapore also showed that attendees were more focused on whether the supply chain can rapidly expand capacity, rather than on any substantial decline in end demand. This means that as AI infrastructure continues to expand, project delivery speed, supply chain assurance and execution capability may become important factors affecting corporate competitiveness and investment returns.

This trend also extends to the field of critical raw materials. UBS's analysis is consistent with the idea of "investing in bottleneck assets," that is, focusing on companies that control key materials and can meet the needs of grid expansion, data center construction, manufacturing reshoring and defense spending, with rare earths, tungsten and other metals being typical examples. As infrastructure investment continues to grow, companies that control the supply of scarce raw materials may attract more investor attention.