Meta, Amazon, Microsoft, Alphabet Seen Spending $750 Billion on Data Centers

Record Capital Spending

Meta Platforms, Amazon.com, Microsoft and Alphabet are expected to spend roughly $750 billion on capital expenditures this year. That figure exceeds the gross domestic product of Ireland or Belgium and is more than half of what all companies in the S&P 500 spent in total last year. Spread across US households, it would amount to about $5,500 each.

The spending is driven primarily by data center construction for AI. John McWilliams of the research group at real estate services firm Cushman & Wakefield said the industry has never seen anything like it, though he added that once every cost category is taken into account, the total no longer seems as crazy as the number might suggest.

Rising Costs Across the Board

Construction materials such as steel and cement have become sharply more expensive due to trade wars, fuel expenses and geopolitical conflicts. The AI boom itself has pushed up prices for items ranging from chips to copper, with the cost of short-term memory used in data centers now more than five times higher than a year ago, according to data compiled by Bloomberg Intelligence.

Demand for power generation and transmission equipment is so high that suppliers report backlogs stretching beyond 2030. Local opposition and regulatory setbacks are also adding to costs, forcing developers to redraw plans and seek costlier alternatives for resources such as power. Oracle issued a force majeure notice on a large New Mexico data center after a key gas pipeline ran into regulatory delays.

$30 Billion Per Gigawatt Site

Cushman & Wakefield provided a cost breakdown for building the shell of a single-gigawatt AI data center in Abilene, Texas, enough to power roughly 750,000 homes. Combined with hardware estimates from Bloomberg Intelligence, a theoretical gigawatt facility in the area could cost a developer roughly $30 billion.

Over the next four years, more than 60 gigawatts of data center capacity is expected to be added in the US alone, tripling current levels in the country that already hosts the most computing power in the world. Similar build-outs are underway globally, from Brazil and China to the UK and Europe, supercharging markets for natural gas turbines, power transformers, construction, real estate, hardware and networking equipment.

The investments rest on the assumption that demand for AI systems will keep growing and that the chips inside these facilities will pay dividends for years to come. Yet some of the world's most powerful AI model developers have yet to turn a profit. Investors including Michael Burry and Bridgewater Associates founder Ray Dalio have warned of an AI bubble, with Burry pointing to circular deals in which hardware makers such as Nvidia partner with and invest in key customers like OpenAI.

Sources: Techmeme