← 資訊首页 | 货源大厅

4.5GW of Natural Gas Will Power AWS's 36-Building, 1,100-Acre Pennsylvania Campus

Projected to employ 1,500-2,000 people, the Homer City Energy Campus pairs hyperscale compute with dedicated gas-fired generation — an architecture built to keep power close to the racks.

One number frames the scale of Amazon's newest compute bet: 4.5 gigawatts of natural gas generation, dedicated to a single campus.

AWS intends to build 36 data centers across a 1,100-acre site in Homer City, Pennsylvania, in a project branded the Homer City Energy Campus. The plan, reported by TechRadar, is expected to create around 1,500-2,000 jobs. Three dozen facilities behind one fence is an unusual configuration; hyperscale campuses more typically add buildings incrementally across multiple phases. Committing the full footprint up front suggests Amazon is planning capacity additions measured in years, not a single construction cycle.

For anyone tracking AI capacity, the gas backbone is the load-bearing detail. Rather than relying on grid supply alone, AWS is backing the campus with 4.5GW of gas generation — power positioned alongside the compute it feeds. The figure matters because it is dedicated to a single site rather than drawn as a share from a regional grid. As training clusters swell past what local utilities can serve, pairing generation directly with the campus has become a defining design choice for gigawatt-class builds.

The choice of natural gas is consequential. Gas plants deliver firm, around-the-clock output, but they tie the campus's operating costs and emissions profile to fuel markets — a trade-off that will draw scrutiny as the project advances.

The jobs figure gives Pennsylvania a clean headline: 1,500 to 2,000 positions attached to the energy campus. The heavier economic weight, though, sits in the equipment bill — turbines, transformers and grid hardware needed to energize 36 buildings at once.

What is not yet public is timing. Land, buildings and 4.5GW of power are pledged; converting that into running IT load is the harder task — and the milestone AI capacity buyers will be watching.


相关阅读


萬安算交所 AIXX · 算力硬件实名会员制交易平台 · 进入货源大厅