Amazon is promising more than $1 billion over the next five years for U.S. communities that host its data centers, pairing the new spending with transparency and utility-cost commitments as the rapid buildout of AI infrastructure runs into growing local resistance.
The initiative, called Built Together, is intended to fund education, job training, energy and water projects and other priorities in places where Amazon operates large computing facilities. Reporting on the program says the company plans to support community-college costs, expand skilled-trades training and fund energy-efficiency work alongside more flexible local grants.
The announcement also includes changes that go beyond community spending. AWS CEO Matt Garman said Amazon is no longer using nondisclosure agreements with government agencies on data-center projects. The company also says it will publish energy and water use annually and structure its power arrangements so local customers’ electricity bills do not rise because of the power required by Amazon data centers.
Those commitments address some of the issues that have made data-center development politically difficult: residents and local officials increasingly want to know what projects will consume, who will pay for grid upgrades and how much information governments can disclose before a campus is approved.
Garman is pushing back on much of the criticism at the same time. He said more than 100 data-center moratoriums are being considered around the country and argued that misinformation is distorting debate about environmental impacts and electricity rates. He warned that broad restrictions could slow U.S. progress in the global AI race.
That is Amazon’s case for faster construction, not evidence that community concerns are unfounded. The Associated Press reported that about six in ten Americans support limiting the number of new data centers that can be built, while a majority are extremely or very concerned about the effect of data centers on electricity prices or local water supplies. Those figures do not establish that every proposed project raises rates or strains water systems, but they show why the infrastructure boom has become a political problem rather than only an engineering one.
The scale of Amazon’s broader spending adds context. The company said in July that it expected roughly $220 billion in capital expenditures in 2026, including data centers and other technology investments. The five-year community commitment is therefore one part of a much larger infrastructure expansion, rather than a substitute for the capital required to build computing capacity and connect it to power.
That distinction is important because community grants cannot by themselves solve the physical constraints behind many data-center disputes. New campuses can require major electrical connections, transmission or generation upgrades, cooling systems, land and permits. Amazon’s promises on power costs, water reporting and local investment are attempts to address the consequences and trust issues around those projects; whether they resolve individual disputes will depend on the conditions in each community.
The decision to stop using NDAs with government agencies may prove as significant politically as the dollar figure. Data-center deals have faced criticism when residents learn about large projects only after negotiations are well advanced. Greater disclosure gives communities more information earlier, although the practical effect will depend on what governments and developers publish before final approvals.
Amazon’s new program is ultimately a bet that the AI infrastructure race needs a stronger local social contract. The company still wants data centers built quickly, and Garman is making an aggressive case against moratoriums. But the combination of community funding, disclosure changes and utility commitments acknowledges a constraint that money spent on chips and servers cannot remove: large-scale AI infrastructure also needs communities willing to host it.
