From the studio

Video: Insights on big tech earnings, NVIDIA’s GTC, and the surge in AI deals (4 min)
Video: Paul Donovan on FOMC and Trump-Xi trade talk takeaways (7 min)
Podcast: What’s next for Chinese equities and the USD? (12 min)

Thought of the day

Dealmaking in the AI space has continued apace. OpenAI and Amazon on Monday announced a seven-year, USD 38bn agreement for the former to access advanced chips on the latter’s cloud infrastructure platform, while Microsoft said it will buy USD 9.7bn worth of computing capacity from Australian data center operator IREN.

These transactions are the latest in a string of AI megadeals unveiled in recent months that have fueled the ongoing stock rally. Alongside the higher capital spending announced by leading tech companies over the past week, these investments underscore the growing need for computing power driven by increasingly complex AI applications, in our view.

We now forecast global AI capex spending to reach USD 423bn this year (from our previous estimate of USD 375bn) and USD 571bn in 2026 (from USD 500bn). By 2030, we expect overall spending to hit USD 1.3tr, implying a 25% compound annual growth rate (CAGR) over the next five years. We see several reasons why these numbers are realistic rather than overly bullish.

Compute demand is outpacing expectations, and monetization is accelerating. Third-quarter earnings and recent company commentary confirm that demand for AI and computational resources remains strong. For example, Google’s Gemini reported a 130-fold increase over the past 18 months in the consumption of AI tokens—the small units of data that large language models use to process and generate output—while Meta said its compute needs have continued to “expand meaningfully” and exceeded its expectations. The anticipated growth of agentic AI—AI systems capable of autonomous decision-making and action—and physical AI, such as robots and autonomous vehicles, should spur greater demand for AI compute. On the other hand, accelerating cloud revenue growth across leading platforms reinforces our confidence in AI’s substantial monetization potential, even relative to the scale of capex plans.

Big tech companies’ margins and balance sheets remain strong despite the capital outlay. The significant upfront investments required for AI compute and infrastructure have nearly doubled capex intensity (capex as a percentage of revenue) for the big 4 US tech companies to 20.8% over the past five years, with expectations to reach 27% by 2030. However, we expect their margins to remain relatively resilient amid a slower growth in other operating costs, which represent nearly 90% of big tech companies’ expenses. These firms also maintain strong cash positions and robust balance sheets. In our view, the perceived risk of underinvesting will likely outweigh that of overinvesting.

AI spending remains modest relative to global GDP. Our AI capex projection of USD 1.3tr by 2030 would account for around 1% of global GDP, according to estimates by the International Monetary Fund. This is below the historical infrastructure investment booms of the past 150 years—including railroads, automotive infrastructure, computers, and telecommunications—which ranged from 1.5% to 4.5% of global GDP. Like those past investments, AI is already driving productivity gains: Adecco’s global survey found an average of one hour saved per day, while Forbes reported 52 minutes. Our calculations indicate that future productivity improvements will be sufficient to justify ongoing AI investments and related depreciation.

So, we maintain our conviction that AI-related stocks should drive equity markets, and we believe that under-allocated investors should add exposure to the theme through a diversified approach.

For more details, refer to Intelligence Weekly #88: A deep dive into our new AI capex projection.