Amazon's $1B PR push collides with new solar data center experiments

On 2 October, Amazon revealed a five-year plan to spend over $1 billion on education, job training, and infrastructure upgrades in communities hosting its data centers. The goal, as outlined in a blog post cited by The Register, is to counter what the company calls "widespread reports" of misinformation seeded by foreign actors to slow US infrastructure development. This initiative follows a pattern of corporate messaging that frames local resistance as an external threat rather than a genuine community concern.

The timing is sharp.

While Amazon spends billions to convince locals that their data centers are harmless, Google is simultaneously testing a radically different approach: putting the data centers themselves in space. NPR reported on 1 October that Google launched "Project Suncatcher," a refrigerator-sized satellite carrying four Tensor Processing Units (TPUs). The prototype aims to gauge how these chips handle the "physical stress of spaceflight" and radiation. If successful, this could lead to orbital data centers powered by solar energy, removing the need for local land, water, or power grids entirely. This divergence in strategy highlights a growing split in the industry over how to manage the physical footprint of AI infrastructure.

The water argument

Amazon's defense of its data centers relies heavily on a comparison with golf courses. The company claims that golf courses use 200 times more water than all of Amazon's data centers combined. The Register, which covered the announcement, noted that this argument only works if people dislike golf courses, which most do not.

Amazon argues its facilities are efficient, using only 13,000 gallons a day, equivalent to 42 US households. However, this figure excludes water used in power generation. Most large-scale power sources, including gas, nuclear, and coal, consume significant amounts of water to turn turbines. By focusing solely on cooling water, Amazon obscures the total water footprint of its operations. Critics argue that this selective accounting ignores the upstream environmental costs that are often borne by local communities in water-stressed regions.

This narrow definition of efficiency is a common tactic in the data center industry. It allows companies to present a favorable number while ignoring the upstream costs of electricity. For residents in water-stressed areas, this distinction is critical. The total volume of water removed from local watersheds is what impacts local ecosystems and communities, not just the water used directly in the cooling towers.

Space as an escape hatch

Google's Project Suncatcher offers a different solution to the power and water problem. By moving compute to orbit, Google can harness the sun's energy directly, without the need for fuel or water. Travis Beals, the lead of Project Suncatcher, told NPR that the sun provides "almost all of the power in our solar system," and this project is about tapping into that resource for AI compute.

The prototype satellite runs a version of Google's Gemma AI model for 15 minutes at a time due to heat management constraints. In the coming years, Google envisions clusters of satellites communicating via lasers, forming space-based data centers. This would eliminate the land use and water consumption that are currently fueling local opposition. The shift from terrestrial to orbital infrastructure represents a fundamental rethinking of where and how AI compute is delivered.

However, space data centers are not without their own challenges. The prototype is a "very minimal test," and scaling this to meet the demand for AI services will require significant engineering breakthroughs. The cost of launching and maintaining satellites is high, and the environmental impact of space debris is a growing concern.

Still, the direction is clear.

As opposition to ground-based data centers grows, tech companies are looking to the sky for answers. Amazon's $1 billion PR push is an attempt to defend the status quo, while Google's satellite launch is a bet on a future where data centers do not need to be in your backyard at all.

The human cost of dissent

Amazon's blog post claims that over 100 data center moratoriums are being considered across the US. The company argues that blocking these projects would put the US at a disadvantage in the global race for AI leadership. This framing treats local concerns as obstacles to be overcome, rather than legitimate issues to be addressed.

The Register's report highlights the tension between corporate interests and community well-being. By investing in "indoctrination" of the public, as the article's headline suggests, Amazon is attempting to shape the narrative in its favor. This is a stark contrast to the approach taken by companies like Google, which are exploring technical solutions to the very problems that are causing the backlash.

The race to build AI infrastructure is accelerating, with demand for compute outpacing supply. Google's CEO Sundar Pichai stated in May that capital expenditures for this year would be between $180 billion and $190 billion. This massive investment is driving the rapid expansion of data centers, both on Earth and in orbit.

As the industry grows, the stakes for local communities are rising. The choice between ground-based data centers and space-based alternatives is no longer just a technical one. It is a question of who bears the environmental and social costs of the AI boom. Amazon is spending $1 billion to keep those costs hidden. Google is spending resources to move them out of sight.

The outcome of this competition will shape the trajectory of data center infrastructure. For now, the dust from the construction sites remains, and the satellites are still in their early test phases. The battle for public opinion is as heated as the processors themselves.