�۶���Ƶ

Video:��Sundeep Gantori on AI adoption's steady growth (3 min)
Podcast:��Investors Club: Can the China AI rally continue? (5 min)
Video:��CIO's Devinda Paranathanthri on putting cash to work (2 min)
Podcast:�� (17 min)

NVIDIA on Monday said it would invest USD 100bn in OpenAI to support its data center buildout and other artificial intelligence (AI) infrastructure. The tie-up between the two leading AI companies boosted investor sentiment further, pushing US equities to fresh record highs for a third consecutive day.

Acc ording to the agreement, NVIDIA will supply OpenAI with advanced chips for data centers with a capacity of at least 10 gigawatts (GW) of power. This is equivalent to the needs of more than 8 million US households.

Without taking any single-name views, the latest headlines not only back our conviction in robust global AI capex growth, but also underscore how the rapid expansion of AI infrastructure is directly translating into substantial power requirements. This reinforces our positive view on the Power and resources theme, and we think investors still underestimate the magnitude and duration of this opportunity.

Capital spending in the power industry could double from today’s level over the next five years. NVIDIA recently suggested that AI data center infrastructure spending globally could reach USD 3-4tr annually by 2030—implying potential electricity demand growth of 170 GW globally and 110 GW in the US, a projection in line with industry forecasts. For context, the 2024 US installed generating capacity was about 1,000 GW. Adding about 10% to existing US capacity at about USD 2,000 per kilowatt (KW) implies an investment of about USD 220bn—globally, the same math comes to USD 340bn. Neither figure includes required transmission grid investment, transformers, breakers, and switch gear, which increases the total required investment potentially by 25-50%. In our view, this means existing 2025 capital spending expectations in the power industry could double by 2030.

Accelerating electricity demand points to a growing need for additional capacity and associated grid infrastructure. AI data center development is not the only pillar supporting rising electricity demand globally. The ongoing electrification of the global economy, industrial reshoring trends, and rising energy demand in the developing world all point to an acceleration in global electricity consumption. In fact, the pace of growth is increasingly overcoming the mitigating effect of efficiency gains that kept a lid on power generation and grid infrastructure needs in recent decades. According to forecasts by the International Energy Agency, global electricity use is slated to grow at around 4% annually through 2027, equivalent to adding more than Japan’s entire consumption each year. This, in our view, points to the need to add new electricity supply and the associated energy infrastructure.

Ongoing constraints in access to power mean opportunities across all sources of energy. Adding capacity and grid infrastructure takes time. A survey of data center developers conducted by Deloitte this year showed that the most significant constraint developers face in the US is access to power. This is also confirmed by the number of temporary and mobile power solution providers that are entering the data center power business, including those traditionally operating in the oil field services space. We expect an “all of the above” strategy to meet growing demand. Finding alternatives, including running existing facilities more, and restarting recently shuttered nuclear capacity will be critical, in addition to new solar, natural gas, and nuclear-generating capacity.

So, with sustained demand and ongoing investment underpinning fundamentals, we expect Power and resources companies to offer durable organic growth in the years to come.