If you are reading this, you are using a data center somewhere.
Likewise if you are a user of YouTube, Netflix, Google, Facebook, email, Slack, or if you share pictures of your pets on your iPhone – you’re a Data center user – so consider before you join that Zoom meeting to ban Data Centers – Zoom operates 17 of them.
The workings of markets have not been repealed.
If the graphs of power consumption projection look incredible and preposterous, well, they might be just that.
For those projects that assume unlimited ability to build “behind the meter” gas generation, there are barriers to the availability of gas.
As one of Jigar Shah’s podcast guests explains, below, building your own generation is easier said than done. Other reporting cautions on the availability of sufficient key components to meet requirements.
For those that do not know Jigar Shah, he is an engineer, an early pioneer in founding a solar company, SunEdison, in 2003. He’s been an investor and leader in the clean energy space, a tireless advocate of climate action, and for 4 years, Director of the (then) powerful Loan Programs Office at the Department of Energy under Joe Biden. So he knows the space.
I’ve posted a couple of his important takes on the the popular concept of Data center development that we are hearing so much about, the projections that Hyperscalers are going to build endless “behind the meter” gas generation that will blow up climate initiatives and ruin communities around the country.
I’m still working my way up the learning curve on these issues, but everything I’m seeing suggests panic is unwarranted, and opportunities for clean energy wins exist.
As the global AI race heats up, there is a huge rush to build data centers fast. There’s no lack of money chasing these projects, with tech giants Alphabet Inc., Amazon.com, Meta Platforms Inc. and Microsoft Corp. committed to spending more than $650 billion this year alone. Yet neither ambition nor capital is enough to materialize all the necessary components.
Almost half of the US data centers planned for this year are expected to be delayed or canceled. One big reason is the shortage of electrical equipment, such as transformers, switchgear and batteries. They are needed not just for powering AI, but also for building out the grid that is seeing increased consumption from electric cars and heat pumps. US manufacturing capacity for these devices cannot keep up with demand, and the scarcity has caused data center builders to rely on imports.
Another under-appreciated factor is the inevitable technological evolution of computing technology that may allow for much reduced energy consumption.
Artificial intelligence is consuming enormous amounts of electricity in the United States. According to the International Energy Agency, AI systems and data centers used about 415 terawatt hours of power in 2024. That accounts for more than 10% of the country’s total electricity production, and demand is projected to double by 2030.
This rapid growth has raised concerns about sustainability. In response, researchers at a School of Engineering have created a proof-of-concept AI system designed to be far more efficient. Their approach could reduce energy use by up to 100 times while also improving performance on tasks.
The research comes from the laboratory of Matthias Scheutz at Tufts University.
The researchers tested their system using the Tower of Hanoi puzzle, a classic problem that requires careful planning.
The neuro-symbolic VLA achieved a 95% success rate, compared with just 34% for standard systems. When given a more complex version of the puzzle that it had not encountered before, the hybrid system still succeeded 78% of the time. Traditional models failed every attempt.
Training time also dropped sharply. The new system learned the task in only 34 minutes, while conventional models required more than a day and a half.
Massive Energy Savings in Training and Use
Energy consumption was reduced dramatically as well. Training the neuro-symbolic model required only 1% of the energy used by a standard VLA system. During operation, it used just 5% of the energy needed by conventional approaches.
==
Finally, there are some examples of Data Centers being done right. One collaboration between Google and a major midwest Utility will provide power with a major buildout of solar energy and battery storage, paid for by the Hyperscaler – keeping rates in check while boosting climate goals.
Alphabet Inc.’s Google is moving ahead with plans for a major data center in Michigan that features a 20-year electricity contract requiring it to cover the full expense of adding a haul of new clean power.
The agreement with DTE Energy Co. underscores the intense competition to connect such facilities to the power grid, as well as community concerns that such projects raise utility bills. The project expects to use as much as one gigawatt of electricity, with service starting in December 2027 and reaching full load in late 2028, DTE wrote in a regulatory filing on Tuesday.
The rush by big technology companies to build massive data centers for their operations and artificial intelligence ambitions has strained the electric grid and raised utility expenses for homes and small businesses in some US regions. The issue has reached the highest levels of US politics, with President Donald Trump calling for tech firms to defray electricity costs.
Google said it’s evaluating a site in Van Buren Township near Detroit. The township’s website currently lists a data center project named “Project Cannoli” that attracted community opposition in January.
Currently, DTE produces most of its electricity from fossil fuels, with about 41% coming from coal, 26% coming from natural gas and the remainder coming mostly from nuclear and wind, according to its website. Google said in its statement that its data center operations will be served by 2.7 gigawatts of new resources, including solar, storage and demand flexibility.
The arrangement that Google and DTE are proposing is unusual. For starters, a 20-year contract — especially one for an amount as massive as one gigawatt — is significantly longer than the typical utility-customer agreement, a fact that DTE acknowledges in testimony filed with regulators. Google’s commitment in turn to fully cover the costs of 1,600 megawatts of new renewable energy and 480 megawatts of battery storage is also novel and may serve as a playbook for other tech firms desperate electricity to power their ever-growing AI computing needs.