Nebius breaks ground on gigawatt factory

AI infrastructure just got a lot more physical. Nebius broke ground on May 12 in Independence, Missouri, for what it calls its first gigawatt-scale AI factory in the U.S. The point is simple — if you want to train models, serve inference, and compete for enterprise AI demand, you need enormous amounts of power, land, cooling, and capital, not just clever software. That is the real news here: another AI company is moving from renting capacity to building industrial-scale compute from the ground up. ### What is Nebius actually building? Nebius is building a multi-building campus in eastern Independence on roughly 400 acres. The company says this is its flagship U.S. AI factory campus, and project materials tied to the city approval process point to potential capacity of up to 1.2 gigawatts. Data Center Dynamics says plans allow for as many as 10 buildings, which gives you a sense of the scale — this is not one warehouse with GPUs in it, but a full campus meant to grow in stages. (businesswire.com) ### Why call it an “AI factory”? Because the industry wants you to think of compute like industrial output. An AI factory is basically a data center campus optimized for training and inference at very high utilization, with power delivery and cooling treated as the main bottlenecks. That framing matters — it tells customers and investors that the scarce thing is no longer just chips. It is the whole stack around the chips: substations, grid access, buildings, networking, and time. (businesswire.com) The factory metaphor is marketing, but it is also pretty literal now. ### What changed this week? The big change is that this moved from approved plan to active construction. Independence’s city council had already approved the Chapter 100 incentive package on March 3, 2026, which let Nebius proceed. Tuesday’s ceremony marked the start of the first phase. So the story is not “Nebius wants to build.” It is “Nebius has started building.” That is a different level of commitment — financially, politically, and operationally. (businesswire.com) ### How big is the economic pitch? Nebius and Missouri officials are selling the project as jobs plus tax revenue. The company says the build should create about 1,200 construction jobs, mostly from local union trades, and around 130 permanent high-tech roles once fully operational. It also says Payments in Lieu of Taxes are projected to deliver more than $650 million over 20 years to the city, school districts, and other local jurisdictions. (nebius.com) Those are the numbers doing the persuasive work. ### So why are people upset? Because hyperscale data centers bring real tradeoffs. Local coverage shows residents have pushed back on power demand, tax incentives, and the broader process around approval. The controversy got big enough to fuel petitions, lawsuits, and even election fallout in Independence after council members backed the incentive package. Nebius has tried to answer that with promises around low water use, noise reduction, ratepayer protections, and community benefits like school support — but turns out those promises do not erase distrust overnight. (businesswire.com) ### Why Missouri? Partly because Nebius already has operations in the Kansas City area. But the deeper reason is utility structure and room to expand. The campus connects to Independence Power & Light, the city-owned utility, and Nebius has emphasized that the project is designed not to raise residential rates. In this market, a site with land, local political support, and a plausible power path is gold. That is why second-tier metros are suddenly in the AI race. (kcur.org) ### What does this say about the AI market? Basically, the center of gravity keeps moving from models to infrastructure. Nebius is also expanding elsewhere — including a 310 MW AI factory in Finland — and it recently raised billions through convertible notes while signing infrastructure deals tied to NVIDIA and Meta. The inference here is straightforward: the company thinks demand for AI compute is durable enough to justify very large, long-dated physical bets. (nebius.com) ### Bottom line? This project matters because it shows what AI competition looks like once the demo phase ends. It looks like power contracts, tax packages, construction crews, and local politics. Nebius broke ground on a Missouri campus, but the bigger story is that AI is turning into heavy industry. (businesswire.com) (nebius.com)