The pushback on Amazon’s data centers has a one-line retort from the man who runs them. “If you go to a community and say, ‘Do you not want to use Netflix?,’ they’ll go, ‘No, no, I still want Netflix,'” Matt Garman, the chief executive of Amazon Web Services, said in an interview released October 10. The point: the services people refuse to give up are the same services his buildings make possible.
Garman’s answer, given on the “A16z Show” podcast, is the latest move in an argument that has turned data centers into a political liability across the United States. Communities in state after state have fought new projects over electricity, water and land use, and Amazon has said more than 100 proposed moratoriums could stall the country’s AI infrastructure buildout.
Polling makes the headwind plain. A Fox News survey last month found 71 percent of registered voters opposed building a data center in their area to support AI development. The cloud industry’s response, so far, has been to spend its way toward goodwill and to argue the national interest, a case Garman has now supplemented with a question about daily habit.
Garman conceded that the company has been too quiet. “I think we need to be more vocal and be more upfront because we actually do a ton that’s really beneficial,” he said. The benefits he named are the ones communities actually weigh: jobs, tax revenue, and upgrades to the local power grid that would otherwise not get built.
The Netflix reference is not idle. Netflix has run the vast majority of its computing on AWS for years, and said in its 2025 annual report that switching to another provider would be difficult. When Garman asks a community whether it wants Netflix, he is really asking whether it wants the cloud that Netflix runs on, with all the machinery that cloud requires.
The stakes are measured in the hundreds of billions. Amazon has said it plans roughly $220 billion in capital spending this year, much of it on data centers, as it races Microsoft and Google to build the capacity that AI models consume. Spending on that scale has collided with local resistance in a growing number of places, and the hyperscale industry’s electricity demand now draws scrutiny nearly everywhere it builds.
Garman’s remarks extend a campaign Amazon began in earnest a week earlier. On October 2, he published a long post warning that data-center moratoriums would set back the largest infrastructure buildout since the interstate highway system, and that the country “could be writing its own losing ticket” in the global AI race. The same day, the company said it would invest more than $1 billion over five years in the communities that host its data centers.
The Netflix line is a change in register from the earlier warnings. The October 2 post argued from national interest and economic consequence. The podcast answer argues from habit, from the contradiction Garman sees in a public that streams, shops and scrolls on the cloud while objecting to the buildings the cloud requires. One is a case to Washington; the other is a case to the neighbors.
The argument has a weakness that critics are quick to press. Communities that oppose data centers often say they are not against the internet, but against a particular facility in a particular place, with its water and power demands, when the benefits accrue to a global corporation. Garman’s reply assumes a clean line from Netflix use to data-center acceptance, and local politics does not always honor that line.
There is a second tension inside his own framing. Netflix is an AWS customer because the cloud offered it scale and resilience that running its own racks did not. But the computing Netflix buys sits in data centers, and those data centers are now the thing neighbors object to. The customer Garman cites as proof of the cloud’s value is also proof of the appetite that keeps demanding more of it.
The stakes for Amazon go beyond public relations. AWS is the profit engine of the company that bears its name, and its ability to keep building data centers determines whether it can keep pace in a race where missing a year of capacity means losing customers to rivals. Every town meeting that delays a project adds months to a timeline that already runs years. Garman’s case rests on a local ledger: construction and operating jobs, property and sales taxes, and grid investments that the data-center tenant finances rather than the local ratepayer.
Whether the charm offensive works is an open question. Amazon cannot outspend its way past every town meeting, and Garman’s own words concede that the industry’s deeper problem has been a failure to explain itself. The company that built its name on logistics now has to sell a harder idea: that the internet people already use depends on infrastructure they keep voting against. The Netflix question is a good line. It remains to be seen whether it can move a room that has already decided.
