I live in Prince William County, Virginia, part of Northern Virginia, the largest data center market in the world. I have also attended a meeting of a local civic group opposed to further data center development. I listened as residents wrestled with what happens when industrial-scale computing moves into communities that never expected to host it.

That debate is no longer confined to county hearing rooms and online virtual forums.

According to CBRE’s 2026 global report, Northern Virginia remains the world’s largest data center market and is operating at nearly full capacity. CBRE says future expansion is becoming harder in Loudoun and Prince William counties because of zoning and approval hurdles.

Yet the more revealing development is what happens when the industry starts looking beyond land.

Elon Musk’s SpaceXAI is preparing to take AI computing into orbit. NVIDIA announced on August 24 that SpaceXAI’s first-generation Starmind satellite will use an optimized version of its Vera Rubin computing system. SpaceX says a new Texas factory could begin producing thousands of AI satellites as soon as late 2027.

Peter Thiel, the billionaire technology investor and co-founder of PayPal and Palantir, is betting on another frontier: the ocean.

In May, Thiel led a $140 million financing round for Panthalassa, a company developing autonomous, wave-powered computing systems that process AI workloads at sea. The company plans pilot deployments in the northern Pacific this year and commercial deployments in 2027.

China has already put computing beneath the sea. The first phase of Shanghai’s wind-powered underwater data center began operating in May, with the full project planned to reach 24 megawatts.

I asked a former AI industry executive whether Musk’s orbital-computing idea was driven by growing resistance to terrestrial data centers and whether it could actually work. He did not attribute Musk’s move to the backlash, but on the technology, he was strikingly confident. Like fusion, quantum computing, and solid-state batteries, he said, the issue is now “when, not if.” Many of the technical problems are already understood; scaling the systems and bringing down their cost remain the larger challenges.

None of this means Virginia’s enormous server farms are about to be replaced by satellites or computers floating in the ocean. Terrestrial data centers will remain the backbone of AI computing for years.

But something important is changing. The geography of computing is becoming part of the contest itself. First, the bottleneck was chips. Then electricity. Now companies must find places where enormous computing facilities can actually be built.

The resistance is substantial. Gallup found in March that 71% of Americans opposed construction of an AI data center in their area. Data Center Watch reports that at least 75 projects worth approximately $130 billion were blocked or delayed by local opposition in the first quarter of 2026 alone.

At the same time, demand is moving sharply in the opposite direction. Lawrence Berkeley National Laboratory estimates that data centers could consume nearly 12% of all U.S. electricity by 2030. CBRE warns that power shortages and construction timelines could constrain new data-center capacity through the end of the decade.

That makes 2026 to 2030 an unusually important window.

The real competition is not Prince William County versus SpaceX, or land versus ocean versus orbit. It is the United States competing with a China that now treats data, computing, and artificial intelligence as elements of national power.

The U.S.-China Economic and Security Review Commission reported this month that Beijing regards data as a strategic resource for economic growth, AI development, intelligence collection, and military power. A separate commission study found that China is coupling rapid AI development with widespread deployment across factories, logistics networks, and robotics.

China’s underwater data center therefore deserves more than a raised eyebrow. It is evidence that Beijing is experimenting with new ways to marry energy and computing capacity while America argues over where the next terrestrial facility should stand.

America retains formidable advantages. We have the leading chip companies, frontier AI laboratories, deep capital markets, and the world’s largest data center market.

But technological leadership is not permanent. Superior algorithms are of limited strategic value if the physical capacity to run them cannot be built fast enough.

What is really at risk is time — and the strategic advantage that time buys.

The danger is not that one rejected project in Virginia hands China the advantage. It is that hundreds of delays, years-long waits for power connections, and harder siting decisions accumulate while our principal competitor treats computing capacity as strategic infrastructure.

America is not starting from scratch. President Trump’s AI Action Plan calls for faster permitting, new power generation and transmission, and greater use of federal land. His administration is turning former uranium-enrichment sites in Ohio and Kentucky into major energy-and-computing hubs, while his Ratepayer Protection Pledge seeks to keep the new infrastructure costs off household electric bills.

Congress is moving as well. Senator Jon Husted (R-Ohio) has proposed legislation to protect ratepayers from data center infrastructure costs, while Virginia Senator Mark Warner (D) has proposed greater disclosure of large data centers’ energy and water demands.

What remains missing is not a commitment to build. It is a clearer strategy for where America should build.

We need a national geography of computing power — not a federal zoning map, but a strategy for matching the biggest facilities with places best able and willing to host them. That means sufficient power, transmission, fiber, land, and, where needed, water.

Some computing must remain near population centers and major fiber networks. The largest facilities will gravitate toward energy-rich regions, while former industrial and federal sites offer another option. Specialized computing may someday move offshore, and certain workloads may eventually make economic sense in orbit.

The larger objective is straightforward: America must build enough computing capacity, in the right places and on the right timetable, to sustain its economy, scientific leadership, and national defense.

That raises a more consequential question than whether every proposed data center should be approved: How much computing power will America need by 2030, where should it live, and will it be ready when we need it?

Christian prudence rejects both technological panic and technological inevitability. Technology does not choose its own destination; people do. That makes the geography of AI not merely an engineering problem, but a human responsibility.

The cloud was never really in the cloud. It has always lived somewhere.

The strategic question now is whether America chooses where its AI future will live — or lets scarcity, delay, and its competitors choose for us