From the studio

Podcast:Signal over Noise with Ulrike Hoffmann-Burchardi on Apple, Spotify (7 min)
Podcast: Monday Jump Start: Anticipating NVIDIA's earnings, the Fed rate-cut outlook, and the latest on gold (6 min)
Podcast: What the US-China AI race means for energy markets (9 min)
Video: Our new AI capex forecast and its implications for the AI story (4 min)

Thought of the day

The rise in AI capex shows few signs of stopping. This quarter, US hyperscalers have unveiled a fresh wave of spending plans: Alphabet now expects USD 91-93bn in 2025 capex, Meta is targeting up to USD 72bn, Microsoft plans to double data center capacity over the next two years, and Amazon projects USD 125bn for 2025, with even higher numbers ahead. Without taking single-company views, we have upgraded our global aggregate AI capex forecast to USD 423bn this year and USD 1.3tr by 2030, representing a 25% compounded annual growth rate (CAGR).

At the same time, access to reliable and affordable power is emerging as a decisive factor for successful AI execution.

Microsoft CEO Satya Nadella said his biggest issue with AI right now is “the ability to get the builds done fast enough close to power,” and that, absent data centers and power infrastructure, his fear is having “a bunch of chips sitting in inventory that I can’t plug in.” Amazon’s CEO echoed the concern, calling power availability a potential bottleneck for future growth. And NVIDIA’s Jensen Huang warned that “China is going to win the AI race” due to its lower energy costs and aggressive subsidies, which make power far more affordable for local tech companies.

We remain bullish on both artificial intelligence platforms and enablers, as well as the power and resources sector that supports them. But it’s increasingly clear that the potential of AI will only be realized if matched with meaningful investments to upgrade energy infrastructure:

US AI ambitions demand major energy investments. The acceleration in US hyperscaler capex has resulted in significant demand growth for electricity and energy storage. While only 25% of projected storage needs are currently met domestically, we see this as a catalyst for a major wave of investment and innovation across US energy infrastructure. Recent regulatory adjustments have ensured continued access to imported energy storage systems (ESS) through 2029, giving US data centers the flexibility to scale rapidly and maintain global leadership in AI infrastructure. Reliance on Chinese battery imports may be only a near-term solution, though. We therefore anticipate that strong US hyperscaler demand will draw in more investments from domestic US energy innovators and providers, capable of meeting megacap tech companies' needs for advanced storage, grid modernization, and new generation capacity.

China’s new energy mix is strategic, not just green. China has accounted for roughly half the global gain in electricity generation over the past 12 months, according to estimates from think tank Ember, with three quarters of this coming from solar. Beyond the environmental benefits, we believe this push into solar (and battery storage) aims to reduce China’s dependence on imported fossil fuels, and to support its domestic tech sector. Subsidies for ESS projects in western Chinese provinces are driving rapid growth, with premium electricity costs at night creating attractive arbitrage opportunities. The “Eastern Data, Western Computing” initiative will connect abundant sources of renewable energy to new data centers, reinforcing China’s AI infrastructure buildout. These policies are designed to create a cost advantage for Chinese tech firms, allowing less efficient domestic AI chips to compete despite higher energy usage. The recent Politburo meeting further confirmed this strategic shift, with China’s energy policy now a clear pillar of its industrial ambitions.

AI and electricity demand are intertwined, but compute is not the only driver. The relentless rise in AI capex is fueling demand across the entire power and resources value chain. But investors shouldn’t mistake the power and resources sector as simply another AI proxy. Our analysis shows the correlation between AI and power and resources is relatively low, potentially making it a valuable portfolio diversifier. Beyond AI, electrification of transport and industry is gathering pace, with the potential for long-term cost advantages for regions that execute well. Electrical equipment and infrastructure companies are reporting accelerated order momentum and expanding order backlogs, while utilities companies describe robust pipelines for new capacity. The investment case for power and resources is broad-based, in our view, encompassing grid modernization, energy storage, and the transition to clean energy.

So, we think the potential of AI increasingly depends on execution in power and resources, opening up an attractive opportunity set for global investors. Rising AI capex is a powerful tailwind for the entire value chain, from electrical equipment and grid infrastructure to energy storage and renewables. With around USD 3tr in annual investment forecast by 2030 across power generation, storage, grids, data centers, and electrified transport and industry, investors should consider diversified and flexible exposure to companies positioned along this value chain, as the AI revolution becomes inseparable from energy innovation