For instance, a recent analysis by Epoch AI of the total training cost of AI models estimated that energy was a marginal part of total cost of AI training and experimentation (less than 6% in the case of all 4 frontier AI models analyzed), and a recent analysis by Dwarkesh Patel and Romeo Dean estimated that power generation represents roughly 7% of a datacenter’s cost.

While closed-circuit cooling systems (i.e. where all of the water is recycled and none of it evaporates)[33] are technically feasible, they are more costly and therefore less common.

At the level of a hyperscale data center cluster, this can translate into requirements of up to 5 and even 10 GW of power, up from 5 MW - a 2,000 fold increase in the span of a decade [4, 11].

They are very geographically concentrated - only 32 countries have data centers, and nearly half of them are in the United States. The state of Virginia has the highest density of data centers globally - it is home to almost 35% of all hyperscale data centers worldwide.

The blog does not detail how the cabinets are connected. Adrian said that, in future, the disaggregated power racks will allow AC inputs to be converted into 400Vdc. Current power solutions convert into 48Vdc, and Adrian argues 400V will be crucial for building more powerful and efficient AI systems. “With 400V we expect improvements and incremental evolution in improved efficiency, like what we have seen in the 48Vdc conversion space,” he said.

ECL launched its first modular hydrogen data center back in May, with a small deployment at its site in Mountain View, California. Each module supports 1MW, and can cool up to 75kW per rack.For the TerraSite, ECL said that it has three pipelines of hydrogen that will feed the facility, with an energy cost of 0.08-0.12/kWh.

Residents rarely learn how data centers may affect their lives until it’s too late. Big tech operators are aggressively deploying nondisclosure agreements to force local officials, construction workers and others to keep these projects under wraps.For tech firms, the incentives to build more of these centers are immense: A McKinsey analysis projected the generative A.I. business could eventually be worth nearly $8 trillion worldwide. Tech companies don’t want to tip off rivals that might try to swoop in and steal a viable site for development. That’s part of the reason they so zealously enforce nondisclosure agreements. But it’s more than that — they also seem to want to avoid angering locals who might learn of the coming disruptions and protest zoning changes.

For starters, if you are seeking to use a green financial instrument to finance your data construction project anywhere in the world, and have European investors, then the Taxonomy Climate Delegated Act (TCDA) will apply. This requires the operator to implement 106 of the EU Code of Conduct for Data Centre (Energy Efficiency) best practices as well as undertake various other activities

decade ago, these plants provided “low carbon” electricity in comparison to the grid at the time, but now in many cases, emit more carbon than local grids. Countries that already have decarbonized grids, France, Sweden, and Scotland, for example, will not benefit from a continuous system that uses natural gas to begin with.

Funneled into the colocation space, this rising demand has almost completely saturated Tier I data center markets like Ashburn, Dallas, and Silicon Valley. As a result, hyperscale workloads are driving growth in Tier II and Tier III markets, occupying existing capacity and spurring new developments.

Addressing the nexus of concerns related to water and electricity consumption — along with the associated impact on carbon dioxide emissions — is one of the biggest challenges faced by any big data center operator. In a blog about its new commitment, Brandt reports that water-cooled data centers use about 10 percent less energy than those using methods related to air cooling. Last year, the company estimated that using water cooling helped Google reduce the "energy-related carbon footprint" of its data centers by about 300,000 tons of CO2. 

Among their chief concerns are the massive amount of stormwater runoff from the millions of square feet of paved over surfaces that would imperil the watershed that provides drinking water for Northern Virginia residents, and the hazardous and noisy construction activities next to residential areas. There are also lingering questions over the impact of energy intensive data centers on the state’s clean energy targets and the true economic benefit for the state.