AI Data Centers Are Becoming the Buyer of First Resort for Hard Tech
"Data Centers are funding the future where no one else will."
Recap
- Opening frame: AI demand turns into budgets for chips, power, buildings, cooling, grid hardware, fiber, and deployment speed.
- Technologies need early buyers: Physical technologies often need production volume, operating experience, and supply-chain maturity before costs fall.
- Capability buyers: NASA and defense historically paid for performance before commercial price efficiency arrived.
- Data centers as meta-alpha product: AI data centers may now be private-sector capability buyers for power, cooling, compute, and speed.
- Public backlash: Local opposition is reacting to real resource conflicts around electricity prices, grid strain, land use, water, noise, tax incentives, and trust.
- Scale evidence: Goldman Sachs estimates roughly $7.6 trillion in AI capital expenditure from 2026 through 2031 across compute, data centers, and power.
- Source scope: The public article cuts off early in the data-center-demand evidence. The unavailable remainder may contain stronger examples or a fuller conclusion.
Context
Not Boring is Packy McCormick's technology and business analysis publication. This piece is an opinionated essay rather than primary reporting. It connects AI infrastructure spending to industrial learning curves, hard-tech financing, and public backlash.
Gallup reported in May 2026 that about seven in ten Americans opposed AI data centers in their local area. Pew's 2026 work found that people often worry about environmental impact, household energy costs, and neighborhood quality of life, while jobs and economic development are viewed more favorably.
Goldman Sachs' Tracking Trillions work estimates about $7.6 trillion of AI-related capital investment from 2026 to 2031 across compute, data centers, and power. The U.S. Department of Energy reported that domestic data-center electricity use rose from 58 terawatt-hours in 2014 to 176 terawatt-hours in 2023 and could reach 325 to 580 terawatt-hours by 2028.
The allocation lens is physical scarcity. AI is bidding for megawatts, transformers, land, construction labor, chips, memory, fiber, cooling systems, permits, and utility planning attention.
Technical Need To Know
- AI data center: A facility built to run dense AI compute, including accelerators, networking, cooling, power, and grid connections.
- Capex: Capital expenditure on durable assets such as chips, servers, substations, power plants, and factories.
- Learning curve: Costs fall and performance improves as production volume and operating experience grow.
- Buyer of capabilities: A customer that pays for what a technology can do before it is cheap.
- Valley of death: The financing gap between promising prototype and scaled commercial product.
- Grid interconnection: The process and bottleneck of connecting new load or generation to the electric grid.
- Firm power: Electricity available when needed, including periods when weather-dependent generation is unavailable.
- Silicon photonics: Using light to move data on or near chips, relevant because AI clusters are constrained by bandwidth, heat, and data movement.
What Folks Are Saying
- Public-opinion data cuts against a simple pro-data-center story: Gallup and Pew show local opposition tied to power bills, environmental impact, neighborhood disruption, land use, water, jobs, and developer trust. Infrastructure data cuts against a simple anti-data-center story: Goldman Sachs, DOE, IEA, and EPRI all point to data centers as meaningful electricity-demand growth. Compute-scarcity stories, including large reported AI infrastructure contracts, make the capability-buyer frame easier to understand.
Nuanced Take
Data centers are becoming a procurement regime. They turn AI demand into hard orders for physical capacity, and those orders can finance technologies that normal markets might leave stranded at pilot scale.
Apollo was publicly funded and mission-driven. AI data centers are privately owned, profit-seeking, and locally sited. Their spillovers may be real, but they are not automatic. If ratepayers absorb grid costs, if gas generation fills the gap, or if communities receive little durable benefit, then learning curves can become private upside and public burden.
The allocation question is who captures the upside. If data-center demand accelerates cleaner firm power, better grid hardware, faster construction, cheaper photonics, and more resilient supply chains, the spillovers could matter beyond AI. If it mostly locks up scarce power and equipment for a narrow set of hyperscalers and labs, the backlash will look much more rational.