Artificial intelligence is no longer only a software race. Stargate made that clear by putting a $500 billion, four-year infrastructure ambition around the compute required to train and serve frontier models in the United States.
OpenAI's launch announcement named SoftBank, OpenAI, Oracle and MGX as initial equity funders and said deployment would begin with $100 billion. The first construction was already underway in Texas, with additional US campuses under evaluation.
The bottleneck has moved into the physical economy
AI capacity depends on land, substations, generation, cooling, fibre, semiconductor supply and permitting. That makes every large compute campus an industrial project with consequences for utilities, construction companies and local labour markets.
The size of Stargate also changes how states compete for technology investment. Tax treatment and incentives matter, but so do power availability, transmission capacity, water constraints and the speed at which large sites can move from land acquisition to energisation.
Announced capital is not the same as completed investment
The $500 billion figure is an intended programme size over four years, not a completed spend. That distinction is essential. Large AI infrastructure programmes typically arrive in phases, and individual campuses can change in timing, ownership structure or capacity as customer demand and power availability develop.
The most useful way to track Stargate is therefore site by site. Land control, power agreements, construction starts, installed hardware and commissioned capacity provide stronger evidence than repeating the programme headline.
Our view: AI has become an industrial-policy sector
American Commerce Review's view is that Stargate belongs in the same analytical category as semiconductor fabs and battery plants: it is private technology investment with public infrastructure consequences.
The winners will not simply be the states that attract an announcement. They will be the regions that can convert promised capital into operating megawatts, local supplier activity and durable high-productivity employment without pushing infrastructure costs onto the rest of the economy.