Robust demand remains key to AI trade despite financing concerns

Podcast: AI’s impact on the economy, Fed, and portfolios, on Apple and Spotify (7 mins)
Video: Three reasons why we find Japanese equities attractive (6 mins)
Video: CIO’s James Cheo on the case for Singapore equities (5 mins)

NVIDIA has partnered with six major financial institutions to assemble a financing package of over USD 500bn to invest in AI infrastructure. According to a statement, the firms will “create dedicated pools of capital at significant scale at attractive rates” for NVIDIA’s customers.

Details on financial terms or the timetable for deployment were unavailable at the time of writing, but the initiative is viewed as part of the chipmaker’s efforts to broaden access to its ecosystem for frontier AI developers, enterprises, and cloud providers. It also raises further questions about AI circular financing, where suppliers help fund purchases of their own products and services.

Without taking any single-security views, we acknowledge the circularity of such arrangements and hyperscalers’ shrinking free cash flows. Concentration risks are also worth noting, given that only a handful of established AI developers currently dominate demand.

But while we think circular financing is a risk to monitor, robust demand, improving monetization, and growth in advance orders for computing power underscore our constructive view on the AI trade.

AI monetization has exceeded expectations. Major US hyperscalers reported average cloud revenue growth of 48% for the second quarter, well above consensus expectations, and a continued acceleration from the 40% growth in the first quarter and 34% in the last three months of 2025. This indicates that AI monetization trends continue to progress well, with cloud demand growing faster than earlier forecasts. The emergence of open-source AI models has also shifted some value creation toward owners of compute and cloud infrastructure, rather than only AI model developers.

Growth in advance orders bodes well for further revenue gains. We expect the acceleration in cloud revenue to continue, forecasting growth of 58% in the current quarter, as hyperscalers reported further increases in advance cloud orders to USD 2.3tr that have yet to be recorded as revenue. While a limited number of AI developers account for nearly half of these commitments, strong demand for cloud computing power should continue to support healthy conversion into top-line growth.

Earnings should stay supported. Given the robust underlying demand for AI compute and cloud, we expect earnings at key AI companies to remain strong, which should continue to drive stock performance. Hyperscalers’ capex beyond 2027 remains a key risk for semiconductor companies, but we believe AI spending of USD 900bn this year and USD 1.2tr next year offers a constructive backdrop over the near to medium term.

So, we think the AI trade has further upside, and believe investors should retain diversified exposure across the value chain. We also see value in more defensive areas of the tech sector, and consider exposure to industrials, utilities, and health care as another way to position for AI-driven growth.