Each quarter, I take a look at the capital expenditure numbers popping out of the main cloud suppliers, and each quarter, they handle to shock me. This time, although, we’ve crossed into territory that’s genuinely arduous to course of. The 5 largest US cloud and AI infrastructure suppliers (Microsoft, Alphabet, Amazon, Meta, and Oracle) have collectively dedicated to spending between $660 billion and $690 billion on capex in 2026, practically doubling what they spent only a 12 months in the past.
Amazon leads the pack with a projected $200 billion, most of it directed at information facilities and AI compute. Alphabet has dedicated $175 to $185 billion, up from $91 billion in 2025. Microsoft is monitoring towards $120 billion or extra, having already spent $37.5 billion in a single quarter. Meta is at $115 to $135 billion, and Oracle is concentrating on roughly $50 billion. For context, 4 corporations alone plan to take a position about $630 billion, a 62% improve from the file $388 billion they spent in 2025.
What are they shopping for? Largely chips, buildings, and electrical energy. The spending splits throughout the AI information heart stack: GPUs and customized accelerators akin to Google’s Tensor Processing Items and Amazon’s Trainium, the bodily information facilities themselves with their liquid cooling and high-density flooring, and infrequently the facility infrastructure to run all of it, together with energy buy agreements and even on-site gas-fired era the place the grid can’t sustain. Roughly two-thirds of the cash goes to compute, and one-third goes to amenities, energy, cooling, and networking, with the facility share rising each quarter.
