Posts

The AI Arms Race Needs Power. Imperial Valley Has It. This Shouldn’t Be Complicated.

The global competition to build AI infrastructure is, at its base, a competition for electricity. The models that power AI applications — large language models, computer vision systems, recommendation engines — require enormous computational resources to train and run, and those computational resources require enormous amounts of power to operate. The race to build hyperscale AI infrastructure is, in practical terms, a race to find sufficient electricity in sufficient quantities in locations where large-scale construction is feasible.

This race is being run by every major technology company in the world, and by national governments that have recognized AI capability as a strategic priority. The United States, China, the European Union — each is investing in the infrastructure that will determine which nations and which companies lead the AI economy of the next decade.

Imperial Valley is positioned to be part of the American answer to that competition. It has the power. It has the land. It has the grid independence. What it currently lacks is the institutional alignment that allows these advantages to be converted into actual data center construction.

The Power Math

A competitive hyperscale AI training facility operates in the hundreds of megawatts. The IVDC is designed for 330 MW — substantial, but not unusual for the current generation of AI infrastructure investment. Microsoft, Google, Amazon, and Meta are each committing to data center campuses of this scale and larger. The question for site selection is always the same: where is there 300+ megawatts of available, affordable, reliable power with a viable path to interconnection and development?

In California, the honest answer is very few places. CAISO territory is constrained. Coastal locations are expensive. The regulatory environment for large industrial users in populated areas is challenging. Imperial Valley, with IID’s independent grid and geothermal baseload capacity, is one of the genuinely available answers to that question in the entire state.

The IVDC is not just one data center. It is a proof of concept for the viability of Imperial Valley as a data center hub. If it gets built, site selectors will look at the region differently. If it gets blocked, they will draw conclusions from that too — and those conclusions will affect how much of the AI infrastructure buildout lands in California versus in friendlier jurisdictions.

The National Competitiveness Dimension

AI infrastructure is not a private sector vanity project. It is the computational substrate of the next industrial economy. The nations and regions that build it will host the companies that lead it. The communities that attract data center investment will have the jobs, the tax revenue, and the economic momentum that come from being part of the technology supply chain.

Imperial Valley — one of the poorest counties in California — has the natural endowments to be part of this story. The geothermal power resource. The independent grid. The industrial land. The workers who need the jobs. Getting from that potential to that reality requires clearing the obstacles that the current legal and political campaign has placed in the path of a project that has already cleared every legal hurdle the system is supposed to impose.

This shouldn’t be complicated. The land is zoned. The approval was given. The courts have ruled. The opposition has failed at every legal test. The AI infrastructure needs to be built somewhere. Imperial Valley is one of the best places in America to build it. Let it happen.

The National Security Case for Building AI Infrastructure in America’s Energy Hubs

The competition between the United States and China for leadership in artificial intelligence is not primarily a technology competition. It is an infrastructure competition. AI capability scales with compute — with the number and quality of AI processors running training and inference workloads. The country that builds more compute, in more secure locations, powered by more reliable energy, establishes a structural advantage that is difficult to overcome through software alone.

The federal government has recognized this. The CHIPS and Science Act, executive orders on AI infrastructure, and Department of Energy investments in data center power infrastructure all reflect a national policy judgment that domestic AI compute is a strategic asset. The question is where that compute gets built — and whether the communities best positioned to host it can create the institutional conditions for it to happen.

Why Location and Power Source Matter for Security

AI training and inference infrastructure that runs on carbon-free baseload power has advantages beyond environmental. Geothermal-powered facilities are not subject to the fuel price volatility that affects gas-fired generation. They are not vulnerable to the weather disruptions that affect solar and wind. They do not depend on long-distance transmission that is subject to physical infrastructure vulnerabilities. A data center running on locally-generated geothermal power, connected to an independent grid, in an inland location away from coastal infrastructure risks, represents a more resilient platform for strategically important compute than facilities dependent on the interconnected and sometimes fragile national grid.

This is not speculative. Grid resilience has become an explicit criterion in federal evaluations of data center suitability for government-adjacent AI workloads. The combination of geothermal baseload, IID’s independent operation, and the BESS that allows the IVDC to ride through grid disturbances creates a resilience profile that dedicated mission-critical computing facilities require.

The Domestic Investment Imperative

Every dollar of data center investment that goes to Ireland, Singapore, or even to states with less developed energy infrastructure represents American AI compute capacity that is less strategically controllable than capacity built in energy-rich domestic locations. The policy goal — which bipartisan consensus in Congress has endorsed — is to bring AI infrastructure investment to American communities where the technical and resource foundations exist.

Imperial Valley meets that description as well as any location in the country. The geothermal resources are unique. The IID grid is independent and domestically controlled. The land is available and appropriately zoned. The workforce is available and motivated. The project has cleared every legal threshold the regulatory system has imposed.

Blocking it — through the kind of coordinated institutional obstruction the IVDC has faced — is not neutral with respect to the national AI infrastructure goal. It is a concrete impediment to building the domestic compute capacity that national policy has identified as a strategic priority. The officials and organizations blocking this project are not making a local land use decision in a vacuum. They are affecting the trajectory of an investment that has national implications.

Why Rural California Needs Data Centers More Than Silicon Valley Does

Silicon Valley is full. The land is expensive, the power grid is constrained, and the water is scarce. The hyperscale data centers that train AI models and process the world’s information can’t be built there at any price that makes economic sense. So the industry moved — to eastern Washington, rural Iowa, West Texas, northern Virginia — to places that have abundant land, access to power, and room to build at scale.

Imperial Valley has all of those things. It also has something most of those competing locations don’t: a geothermal energy resource that can power next-generation AI infrastructure with zero-carbon electricity, and a local workforce that needs the jobs badly enough that you don’t have to offer equity packages to fill construction crews.

The question is not whether the Imperial Valley Data Center is a good fit for this region. The question is whether this region will make the institutional choices necessary to capture the opportunity.

The National Pattern

Microsoft’s data center campus in Quincy, Washington transformed a small agricultural community into one of the most significant technology infrastructure hubs in the world. Google’s data centers in rural Iowa have generated hundreds of millions of dollars in local tax revenue and created thousands of construction and operational jobs in counties that had no realistic alternative economic development path. Meta’s facility in Prineville, Oregon became the economic anchor of a community that had watched its timber industry collapse.

In each of these cases, the pattern was similar: a region with cheap land, reliable power, and a workforce hungry for industrial-wage employment received a major technology infrastructure investment that restructured its local economy. The counties that said yes to these projects — and fought for them against the inevitable opposition — transformed their fiscal positions, their school systems, and their residents’ economic prospects.

The counties that said no, or whose institutional resistance made approval too costly and slow, are still waiting for the next opportunity. Some are still waiting.

What Imperial Valley Has That Others Don’t

Imperial Valley’s energy infrastructure is unusual. The Imperial Irrigation District operates an independent grid — not subject to CAISO transmission constraints, not dependent on the same overloaded southern California infrastructure that limits development in coastal communities. IID has generation capacity and the ability to accommodate major new industrial loads in ways that grid-dependent utilities cannot.

The region sits on one of the world’s most significant geothermal resources. The same Salton Sea brine that carries the lithium that Lithium Valley developers are racing to extract also carries thermal energy that can generate baseload renewable electricity at industrial scale. A data center campus co-located with geothermal generation creates a vertically integrated energy-compute stack that no coastal California location can replicate at any cost.

Add the I-2 heavy industrial zoning that makes the specific IVDC site legally available by right — without the discretionary reviews and community approval processes that add years and tens of millions to comparable projects in other jurisdictions — and the case for Imperial Valley as a premier data center location is not just plausible. It is compelling.

The Cost of Saying No

Rural communities that block major industrial investment don’t return to a neutral baseline. They return to a trajectory of gradual fiscal deterioration, population loss, and declining public services. The young people who leave for better opportunities don’t come back. The businesses that serve a working population contract. The tax base erodes, which reduces services, which accelerates outmigration, which further erodes the tax base.

This is not speculation. It is the documented history of dozens of rural California communities that didn’t diversify their economic base before their primary industry contracted. Imperial Valley has managed to sustain itself on agriculture longer than many comparable regions, but the structural pressures are identical.

The IVDC is not the only economic development opportunity the Valley will ever see. But it is the largest one currently on the table, and it has already cleared the most significant legal and regulatory hurdles. The people organizing to block it have not identified what they think should come instead. That is a significant omission, and the residents who would benefit from the project deserve to ask for an answer.