Massive AI capex surge

The AI infrastructure buildout drove an estimated $437 billion in datacenter capital spending in 2025 and is forecast to jump another 40% in 2026, making this one of the largest capex cycles in history. The same breakdown flags power and energy as emerging bottlenecks for hyperscale AI capacity, forcing design tradeoffs across servers, cooling and site selection. (x.com)

The companies building artificial intelligence data centers spent about $437 billion in 2025, and one 2026 estimate puts that total above $700 billion. (creativestrategies.com) Creative Strategies said hyperscaler capital expenditure could top $700 billion in 2026, up from roughly $437 billion in 2025. Dell’Oro Group separately projected worldwide data center capital spending would rise more than 30% in 2025 as cloud giants expanded for artificial intelligence. (creativestrategies.com) (delloro.com) The bill is being set by a small group of companies. Meta spent $72.22 billion on capital expenditures in 2025 and told investors on January 28, 2026 that 2026 spending would rise to $115 billion to $135 billion, while Amazon said in February it expected roughly $200 billion of 2026 capital spending, mostly for artificial intelligence infrastructure. (investor.atmeta.com) (cnbc.com) Alphabet reported $91.4 billion of capital expenditures in 2025, with 60% going to servers and 40% to data centers and networking, and told investors on February 4, 2026 that 2026 capital spending could reach $175 billion to $185 billion. Microsoft said on January 3, 2025 that it planned to spend $80 billion in its 2025 fiscal year building data centers for artificial intelligence workloads. (fool.com) (cnbc.com) A data center is a warehouse full of servers, networking gear, and cooling equipment. Artificial intelligence pushes those buildings harder because training and running models require dense clusters of graphics processing units, which draw far more electricity and throw off far more heat than older cloud workloads. (goldmansachs.com) (jll.com) That is shifting what gets built inside the building. Creative Strategies said the spending mix has moved from a historical 50-50 split between equipment and facilities to roughly 66-34 in favor of silicon, while JLL said rack densities are approaching 100 kilowatts and liquid cooling is becoming a requirement. (creativestrategies.com) (jll.com) The bigger constraint is no longer just money. Bloom Energy said in its January 2026 survey that power availability had become “a defining boundary” on data center growth, with longer grid interconnection timelines pushing developers toward power-advantaged regions and on-site generation. (bloomenergy.com) JLL said power, not land price or location, is becoming the main site-selection test because grid connections can take years. Bloom Energy said Texas is positioned to become the largest United States data center market within three years, while older hubs such as California and Oregon are expected to lose relative share. (jll.com) (bloomenergy.com) The spending wave is already showing up in the broader economy. S&P Global said data center and related high-tech investment made United States gross domestic product about 0.5 percentage point larger in the second quarter of 2025 than it would have been if that spending had stayed on its 2011 to 2022 trend. (spglobal.com) The next fight is whether utilities, equipment makers, and local governments can keep up with the orders already on the books. Creative Strategies said many gigawatt-scale artificial intelligence campuses are only now reaching initial operating targets, which means the capex surge is still in its build phase, not its endgame. (creativestrategies.com)

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.