1GW AI data center in Abilene estimated at $30 billion
amazon chips google meta microsoft
| Source: Techmeme | Original article
Building a 1GW AI data center in Abilene, Texas could cost about $30 billion, with $17 billion for chips, $3 billion for power and $2.4 billion for construction.
Bloomberg’s latest analysis puts a concrete price tag on the scale of AI infrastructure that tech giants are racing to build. A 1‑gigawatt (GW) AI data centre in Abilene, Texas, would cost roughly $30 billion to construct and equip. The bulk of the spend – about $17 billion – would go on specialised AI chips, while power infrastructure would require an estimated $3 billion and construction costs another $2.4 billion.
The numbers underscore how quickly AI compute is translating into massive capital projects. A single gigawatt of AI‑focused compute power rivals the output of a large utility plant, and the associated chip demand is reshaping the semiconductor supply chain. At the same time, the $3 billion power budget highlights the growing strain on regional electricity grids, a factor already linked to a surge in gas‑fired generation as we reported earlier this week in our piece on the AI data‑center boom and its impact on smog‑producing power plants.
Why it matters is twofold. First, the sheer scale of investment signals that the leading cloud providers – Meta, Amazon, Microsoft and Google – are prepared to pour tens of billions into dedicated AI capacity, potentially reshaping the geography of high‑performance computing in the United States. Second, the cost breakdown points to where supply‑chain bottlenecks could emerge, especially in advanced AI chips and reliable, low‑carbon power.
What to watch next are the financing arrangements and site‑selection decisions that will follow. Investors will be keen on how companies secure the $30 billion spend, while regulators and local authorities will scrutinise the power‑grid upgrades and environmental permits required for such a facility. The pace at which similar projects materialise will be a barometer for the broader AI race and its downstream effects on energy demand and chip manufacturing.
Sources
Back to AIPULSEN