AWS’s generations-long warning about data-center pauses remains unproven

The word moratorium does not establish whether a proposal would impose a temporary pause, a broad restriction or something else. The available sources do not provide the proposals’ text, duration, geographic reach, exemptions or status. They also do not show which projects would be affected.

Those details matter. A pause affecting one project could have a different effect from a restriction covering several planned facilities. Without proposal-by-proposal evidence, the number alone cannot establish how much construction would be delayed or whether other sites could provide needed capacity.

That uncertainty does not prove that local restrictions would be harmless. It means each proposal should be assessed on its terms, rather than treating an unverified tally as one national policy with a uniform effect.

The investment forecast does not measure losses

The strongest case for Garman’s concern is that data centers supply computing capacity, and a delay to a needed facility could constrain AI development if alternatives were unavailable. That is a plausible chain of events. The evidence here does not show that it applies to every proposal in the count, or that a resulting constraint would last for generations.

Brookings summarized an estimate that investment in AI infrastructure—including data-center buildings, power systems, networking and specialized computing equipment—could total $10.3 trillion from 2025 through 2032. That is a projection for a broad category of infrastructure, not money already spent or an estimate of investment that moratoriums would block. It signals the anticipated scale of the buildout, not the cost of any particular local pause.

Garman argues that data centers matter to U.S. economic and national-security interests. To establish his larger warning, an analysis would need to identify the projects each proposal might delay, estimate the duration and scale of those delays, and connect them to measurable changes in computing capacity, AI deployment or the economy. It would also need to explain why any competitive setback would persist for generations. The materials cited here do not provide that analysis or an independent evaluation of the forecast.

Transparency is not proof of impact

Amazon has announced more than $1 billion in additional community investments over five years through its Built Together program, as GeekWire reported. That is a company commitment, not evidence that the money has been spent or that planned benefits have reached host communities.

Amazon has also said it will stop using nondisclosure agreements with government agencies on its data-center projects, a commitment covered in Tom’s Hardware’s report. Greater access to project information can help local review, but a transparency pledge does not independently establish a facility’s effects on local resources or costs. Its value depends on what information is published and whether it lets communities assess specific proposals.

The choice need not be between approving every project and stopping construction everywhere. Rules about disclosure, who pays for infrastructure or when construction can proceed may have different effects; those effects should be assessed rather than assumed. Amazon’s forecast does not establish that local limits would cause lasting national harm, just as the existence of community concerns does not establish the impact of a specific facility.

The verdict on Garman’s generations-long warning is unproven. AWS and public officials should identify the proposals and projects at issue, and independent analysts should estimate likely delays and their consequences. Until that evidence exists, the warning remains a possible scenario, not a demonstrated reason to dismiss local limits.