Big Tech’s electricity addiction is reshaping our grid, and the public pays the price.
Big tech firms and AI giants love to talk about innovation, the cloud, and digital futures. But behind the buzz lies a grounded reality: a relentless demand for electricity, water, and grid infrastructure that is reshaping entire communities and leaving everyday people to pick up the tab. California is on the frontline of this shift, especially in inland regions like Lancaster, where new data center campuses threaten to outstrip local systems and undermine the state’s clean energy goals.
When you hear the word “cloud,” you might imagine something intangible, floating in cyberspace. But the truth is this: AI and data centers are intensely material. They consume vast amounts of electricity, generate prodigious heat, require water cooling, demand large parcels of land, and need constant reliability.
According to the International Energy Agency (IEA), global electricity consumption of data centers is currently around 415 terawatt hours (TWh), about 1.5 percent of global electricity use. The IEA projects that by 2030 this could nearly double to around 945 TWh, or just under 3 percent of global demand.
In the United States these demands are already being felt. A recent estimate shows that U.S. data centers consumed about 4.4 percent of total U.S. electricity in 2023, and could rise to 6.7 to 12 percent by 2028.
More ominously, the pace of increase is accelerating because of AI. The IEA finds that servers equipped for AI (accelerated servers) are projected to grow in power consumption by about 30 percent annually, compared to about 9 percent for conventional servers.
In practical terms, many modern hyperscale data centers now draw as much power as a small city. And in areas where many are clustered, the local grid is under serious strain.
It may seem odd that utilities, and sometimes regulators, are so eager to welcome new data centers into their regions. But the reasons are straightforward once you pull back the curtain.
A hyperscale data center provides huge, steady demand 24 hours a day, year after year. That is gold for utility business models that benefit from large, stable loads. A data center is far easier to plan around than many residential or small commercial loads, which can fluctuate.
Large loads give utilities justification to build new transmission lines, substations, transformers and sometimes even new generation capacity. These investments bring returns and regulatory approvals. The slow transition to distributed energy and smaller loads often offers lower returns.
Tech companies often secure favorable deals: low rates, guaranteed access, and preferred interconnection. Meanwhile the rest of the grid pays more of the fixed costs. Utilities’ upgrade costs end up in rate increases for residential customers.
Big tech is eager to brand their operations as “green” or “100 percent renewable.” But most data centers are still powered from grids dominated by fossil generation, especially when demand is constant and renewable supply is variable.
When you put all these together, it becomes clear: many utilities see data centers as a win, a high value customer that helps them expand infrastructure. But the burden of cost, risk, and environmental consequence is largely shifted onto ratepayers and communities.
Turning from global and national data to local reality, the city of Lancaster, California in the Antelope Valley north of Los Angeles is emerging as a hotspot for the next wave of data center expansion. While specific project details are sometimes proprietary, the general trend is clear: data centers are targeting California’s inland regions, where land is cheaper and power access more flexible.
Lancaster already faces resource stresses: desert heat, limited local generation, vulnerable transmission access, and water constraints. A new hyperscale data center campus in such a location carries multiple risks: large incremental power draw, heightened cooling demand in hot months, and water use pressure.
Even if the center claims carbon neutral status via renewable energy credits, the physical demand still hits the local grid. When that grid is already constrained and infrastructure upgrades are needed, someone pays. Usually that someone is the local community and ratepayers.
If utility upgrades are required such as new substations, more transmission lines, or backup generation, they are often financed across all customers. The tech company benefits, but local residents bear the risk and cost.
We can and should ask: Is this consistent with California’s ambitious clean energy, climate justice, and resource equity goals, or is the state quietly reshaping its resources to accommodate corporate IT campuses rather than public need?
Focusing only on electricity tells only half the story. These data centers generate intense heat, and once you add cooling and water usage, the resource burden deepens.
According to the Pew Research summary of IEA data, in 2023 U.S. data centers consumed about 17 billion gallons of water directly. That figure is expected to rise, particularly for large AI driven centers.
Cooling systems alone can represent 7 percent to over 30 percent of a data center’s power use depending on the design and location. In arid regions like the American Southwest, the water burden is real and immediate.
For a place like Lancaster, which already faces desert conditions and resource constraints, that is a major concern.
Tech company statements are full of slogans like “100 percent renewable energy” or “carbon neutral data center operations.” But these claims can mask inconvenient truths.
Purchasing renewable energy credits or offsets does not mean that the actual electrons running the servers come from wind or solar. Most large scale data centers are still served by grids where fossil fuel generation dominates, especially at night, during peaks, or when backup is needed. U.S. data centers still draw over 40 percent of their electricity from natural gas according to IEA data.
When you have a massive 24 hours a day load, you need generation you can rely on. That means dispatchable plants such as gas, nuclear, or hydro rather than intermittent wind and solar. The need for redundancy and uptime keeps fossil based infrastructure central despite the green labeling.
Even if the center itself claims clean operation, the wider system impact remains. New loads require new transmission, new generation, and backup power. All of that can lock in more fossil generation, delay plant retirements, or postpone community based renewables.
The “green AI” story often becomes a gloss over a much more resource intensive reality.
There is a resource and justice dimension to all of this. As tech firms expand in places like Lancaster, the consequences are visible.
In California, these patterns throw into relief the tension between clean energy goals and the needs of large scale AI infrastructure. If the system is shaped to meet private corporate demand rather than community equity, the promise of a just transition is undermined.
If current trends continue, the implications are substantial.
These risks call into question whether the AI and data center boom aligns with a sustainable, equitable, clean energy future, or whether it constitutes another wave of utility expansion and fossil fuel perpetuation under a tech banner.
To shift course several things are needed.
The AI boom and data center expansion carry real promise for innovation, digital infrastructure, and economic activity. But they also carry hidden strings: huge demand for electricity and water, large infrastructure costs, and the risk that communities and ratepayers pick up the tab so corporations can reap the benefit.
In California, where energy, water, and climate justice are central public goals, the rise of hyperscale data centers must be treated not as a neutral force but as a major public policy moment. Will the state bend its resources to serve a handful of massive tech campuses, or will it insist that technological growth align with equity, sustainability, and public benefit? The choice matters, and for Lancaster and other inland communities, the consequences are already beginning to arrive.
Sources and further reading