- AWS CEO looks to debunk common ‘myths’ about data centers
- Amazon announces it will no longer make local governments sign NDAs when proposing data center projects
- The company has also laid out new codified community commitments
CEO of Amazon Web Services Matt Garman has released a statement in defense of AI data centers that labels the technology as “The race our nation can’t afford to lose.”
Within it, Garman calls on the United States to band together against “misinformation and outright lies” that are being spread about AI data centers to prevent the country from missing out on the huge benefits that will be provided to the economy, national security, and local communities.
“A lapse in our nation’s focus or resolve could put us behind, and potentially irreparably so,” Garman says.
Debunking the myths
There has been a significant amount of messaging about the positives and negatives of the AI data center buildout, with falsehoods and misinformation fueling both sides.
Garman’s statement starts with a section aimed at countering the ‘myths’ spread about AI data centers, such as water use, higher energy costs, and pollution – but keep in mind that Garman makes no distinction between AI data centers and regular data centers.
For example, Garman argues that data centers account for “0.5% of total U.S. direct industrial water use”, but draws on the figures of direct water use to argue this point.
The indirect water usage of data centers is far higher, and often isn’t included in water estimates for AI data centers. The indirect water usage includes upstream factors such as the water used to generate the huge amounts of electricity that data centers rely on. Data centers are also being built in regions that have experienced long-term droughts, exacerbating water shortages and their effects on local ecosystems.
Garman also labels the rise in US energy bills caused by data centers as “misleading and convenient scapegoating,” arguing that “the accelerated adoption of AI is precipitating a build-out that was destined to happen.” What Garman fails to mention is that the infrastructural build-out is being fueled by AI data centers.
Up until very recently, new grid connections for data centers haven’t been paid outright by the developers. Instead, grid operators have funded new connections and upgrades to power lines through increased bills on rate-payers. It is only the backlash from rate payers that has preceded the introduction of Trump’s voluntary Ratepayer Protection Pledge – and the much more binding (and bipartisan) Ratepayer Protection Act. Had there been no public backlash, these measures would not have been necessary.
The next myth Garman attempts to counter relates to the pollution caused by “large backup diesel generators”. Garman states that these are necessary as the services “consumers, businesses, and governments rely on every day” are provided by data centers. This is, in its exact wording, true.
But the pollution many of those in the US are arguing against is attributed to the fossil fuel burning gas turbine engines that many AI data centers use as an alternative to connecting to regional power grids – particularly if their connection requests are denied. The turbines at some campuses give off the same noise pollution as an international airport, 24 hours a day, seven days a week. The sounds they produce have also been involved in causing neurological symptoms in local residents.
The demand for gas, pushed higher by AI data centers, could see the US use more natural gas than Germany and Japan combined by 2035, and the associated pollution could cause generational health decline and cost $25 billion in associated costs.
The final myth that Garman seeks to debunk is that ‘Community members get nothing from data centers’, arguing that data center operators are “often the single largest taxpayer in a community” and that this “reduces the taxes paid by families, but it also funds investments in schools, fire and police departments, parks, roads, and other community services.”
But data center operators are also the largest beneficiaries of tax cuts and credits too. Numerous states have begun repealing tax cuts for data centers after seeing billions lost in expected revenue – billions that otherwise could have been used for funding investment in community services.
Garman also points to the boom in construction jobs associated with the AI data center buildout, but doesn’t include that many of these jobs are temporary. Once the construction of a data center is complete, these jobs often disappear unless there is continuing demand within the local region, and operational data centers themselves require a very small number of employees.
In an effort to get local communities onboard with data centers, Garman’s statement also lays out a number of ‘codified’ community commitments that Amazon will adhere to.
These commitments fall under four pillars; Protecting & Enhancing Communities; Providing Jobs & Economic Benefit; Engaging Communities; and Building & Operating Responsibly.
The full list of commitments is available here, but there are a few key points to look at. Outside of vague commitments to covering the cost of grid connections and infrastructural upgrades, paying their fair share of taxes, and paying the workers they hire, Amazon will contribute “$1 billion over the next five years” to jobs and education, energy affordability and water solutions, and community investment.
Given that Amazon operates 236 data centers that are operational or under construction, this $1 billion figure amounts to around $4,237,000 per data center over the full five years, or about $847,000 per data center per year.
Amazon has also announced a commitment to “no longer use nondisclosure agreements with the government agencies we work with on our projects.”
This is by far the biggest announcement included in Garman’s statement, yet it takes up just a single line. Local communities have long called out for more transparency around AI data centers.
In numerous cases, NDAs have been used to prevent local governments from discussing planned data centers with their constituents. In numerous other cases, data centers have only been announced to local communities after the necessary permits are secured, preventing local residents from being involved in the consultation process and limiting their ability to oppose new projects.
The community commitments offered by big tech companies such as Amazon and Meta looking to build AI data centers are not out of the goodness of their hearts. They are a direct result of the community backlash against AI data centers that has occurred across the US.
Had this huge opposition not happened these benefits would not be offered, but they are being jazzed up by big tech companies as a benefit they are providing to communities rather than the efforts of grassroots movements and everyday Americans fed up with rising bills, increased noise pollution, and an economic system that favors those with money and power.
