2024: The Year Infrastructure Became the AI Battleground

When historians look back at the AI investment cycle, 2024 may be remembered as the year the infrastructure thesis became undeniable. The combined capital expenditure of Microsoft, Google, Amazon, and Meta exceeded US$200 billion — a figure that would have seemed fantastical three years ago. Data centre construction backlogs stretched to three years in some markets. Power purchase agreements for nuclear energy, once a novelty, became a standard tool in the hyperscaler procurement toolkit. The infrastructure layer of AI was no longer a specialist investment thesis; it was a consensus trade.

The most significant development of the year was the entry of sovereign wealth funds at scale. GIC, Temasek, ADIA, and Mubadala each made substantial commitments to digital infrastructure platforms in 2024, either through direct investments or co-investments alongside specialist managers. This shift matters not just for the capital it brings but for what it signals: when sovereign funds with multi-decade investment horizons make a structural allocation to an asset class, they are expressing a view about the durability of the underlying demand. Their conviction in AI infrastructure is well-founded.

The year also brought the first serious signs of market differentiation. Not all data centre markets are equal, and not all operators are equal within them. Assets in power-constrained markets — Singapore, Northern Virginia, Dublin — commanded significant premiums. Operators with long-term power agreements and pre-leased capacity traded at multiples that would have seemed excessive two years ago. Meanwhile, speculative development in markets without clear demand anchors began to attract scepticism. The market is maturing, and with maturity comes selectivity.

For private investors, the lesson of 2024 is that the window for building AI infrastructure exposure at reasonable prices is narrowing. The assets that were available at infrastructure-like multiples in 2022 and 2023 are now priced to reflect their AI optionality. The opportunity going forward lies in the markets and asset types that have not yet been fully discovered — tier-two geographies, infrastructure adjacencies, and the credit layer that finances construction. At Lensbridge, that is precisely where we are focused.