Trending Now
We have updated our Privacy Policy and Terms of Use for Eurasia Group and its affiliates, including GZERO Media, to clarify the types of data we collect, how we collect it, how we use data and with whom we share data. By using our website you consent to our Terms and Conditions and Privacy Policy, including the transfer of your personal data to the United States from your country of residence, and our use of cookies described in our Cookie Policy.
{{ subpage.title }}
3D illustration of a robot hand reaching out to touch a lightbulb.
The AI energy crisis looms
According to a CNBC analysis of US government data, a single data center operating at 85% capacity consumes as much electricity as 710,000 households or 1.8 million people. There are currently 3,000 data centers across the US, by one estimate, with the greatest number in Virginia (477), Texas (291), and California (285). With artificial intelligence as a leading factor, power demand from data centers is expected to increase 160% by 2030, according to Goldman Sachs. And major tech companies such as Google and Microsoft have revised their environmental goals because of their AI ambitions.
The incoming Donald Trump administration promises to take a deregulatory approach across the board. Lee Zeldin, a former congressman Trump has tapped to lead the Environmental Protection Agency, said he wants to “make America the AI capital of the world.”
So all signals point to Biden’s climate goals, soon in the rearview mirror, slipping further out of grasp.
Startup Cerebras System's new AI supercomputer Andromeda is seen at a data center in Santa Clara, California, U.S. October 2022.
Hard Numbers: Electricity drain, Coal in demand, Ignoring AI, Deal for Palantir, China’s chip fund
9.1: The nonprofit Electric Power Research Institute estimates that data centers will drain up to 9.1% of US electricity by 2030. Last year it was just 4%, but the rise of artificial intelligence has placed newfound demands for easily accessible computing power.
54: The increased energy demands from AI have even slowed US plans to close coal plants. 54 gigawatts of coal-based power generators are expected to be retired by 2030, a number that has fallen 40% from last year’s estimate from S&P Global Commodity Insights.
2: Only 2% of Brits say they use ChatGPT or another AI technology every single day, according to a new survey from Reuters Institute and Oxford University. “Large parts of the public are not particularly interested in generative AI, and 30% of people in the UK say they have not heard of any of the most prominent products, including ChatGPT,” the report’s lead author said.
480 million: Palantir won a $480 million deal with the US Army for a computer vision project. The Peter Thiel-founded company already works extensively with the military and has worked with allied militaries, including Ukraine’s in the war against Russia.
47.5 billion: In the face of stringent US export controls that limit China’s ability to gain access to important semiconductors, the Chinese government announced its third chip fund after similar investments in 2014 and 2019. This fund is a $47.5 billion investment into chip companies, aimed at getting a stronger foothold on the chips necessary for training and running AI models.Joe Biden starts to campaign on AI
On May 8, Joe Biden spoke at Gateway Technical College in Racine, Wisconsin. The president was bragging.
Six years after his predecessor, Donald Trump, visited the same city to boast of Taiwanese tech company Foxconn’s $10 billion plan to bring a LCD manufacturing plant to Racine — that never materialized — Biden chose the same site for a new high-tech manufacturing project of his own. Microsoft will invest $3.3 billion to build a new data center to support artificial intelligence, a project that the company says will bring 2,000 permanent jobs and 2,300 union construction jobs to Wisconsin.
It’s good business, and better politics. Wisconsin is an important swing state for Biden in his forthcoming election against Trump. This latest announcement seemed to mark a moment where Biden accepted that AI is going to be an important part of his presidential legacy — and that it’s a record he should run on.
Right place, right time
OpenAI ushered in the generative AI revolution with ChatGPT midway through Biden’s first term. Silicon Valley rushed to develop it, Wall Street rushed to fund it, and governments around the world rushed to regulate it. Biden was in just the right position to reap the political rewards.
The US hasn’t passed comprehensive regulation to rein in AI, lagging behind its European counterparts in that regard, because it would require Congressional action. Instead, Biden secured voluntary commitments from the top AI companies to reduce the risks of their technology and issued a sweeping executive order dictating that every federal agency and department needs to assess and mitigate the risks AI poses, and how they can safely use it.
Beyond that, AI has become a focus of Biden’s industrial policy and export control measures, both of which have major implications for foreign policy and national security. Microsoft's investment also comes mere weeks after the Biden administration helped orchestrate the PC giant’s $1.5 billion investment in the Emirati tech giant G42, which pledged to restrict ties with China in favor of working with US tech firms.
Federal dollars pour into AI
The Microsoft data center was one in a series of chest-pounding announcements from the Biden administration, which has used funds from the CHIPS and Science Act to incentivize tech infrastructure firms to build in the United States. Taiwan Semiconductor Manufacturing Company will get $6.6 billion to invest a total of $65 billion to expand its chip fabrication complex in Phoenix, Arizona. Samsung will get $6.4 billion to pour $45 billion into its Texas facilities, and Intel will be granted $8.5 billion to construct and expand facilities in Arizona, Ohio, New Mexico, and Oregon.
AI wasn’t necessarily top of mind when the CHIPS Act passed in 2022, said Scott Bade, a senior analyst in Eurasia Group’s geo-technology practice, but it’s become the focus of the government’s efforts to nationalize chips and data centers.
“If you look at the political motivations for the Chips Act, a big part of that was the auto industry not having access to chips during the pandemic,” Bade said. Most of those were so-called legacy chips, not the high-powered graphics processors needed for AI, but the investments and legislation was already in place by the time AI became the hot topic in consumer and military tech.
The US has an advantage over rival China when it comes to artificial intelligence technology, but also the chips and chip-making facilities necessary to train and run powerful AI applications. Not only are many of the most important AI chipmakers — such as Nvidia, AMD, and Intel — American firms, but important non-US infrastructure firms are subject to US export controls because they rely on small parts made in America. The Biden administration has ramped up export controls to give the US an economic and technological advantage over China. And don’t forget the military side — global powers are looking to AI to super-charge their weaponry.
The election looms
AI lets Biden make some important claims in his rematch against Trump, including:
- American companies are leading the world on AI
- Multinational firms are investing in US facilities
- They’re bringing high-tech manufacturing jobs to the US
- And the US is keeping China at bay in the AI space
Not all of those arguments will resonate in retail politics, but Arizona and Wisconsin, where new facilities are popping up, are key swing states looking for good union jobs. In Wisconsin especially, Biden will make the case that he’s delivering what Trump couldn’t.
“The fact that you have a fab in a major swing state that helped him win last time and also has an important Senate race — that's not a coincidence,” Bade noted.
Speaking in Wisconsin, Biden barely mentioned technology, let alone artificial intelligence. Instead, he focused on delivering where Trump could not.
“During the previous administration, my predecessor made promises, which he broke more than kept, left a lot of people behind in communities like Racine,” Biden said.
Artificial intelligence might not be the snazziest talking point for retail politics, but it’s bound to be a major undercurrent — even when it’s not mentioned explicitly.
LEDs light up in a server rack in a data center.
Electric Company and Water Works
The artificial intelligence boom has brought new strains on the environment, namely through demands on electricity and water. The International Energy Agency estimates that, by 2026, the rise of AI and cryptocurrency will lead to an increase in electricity used by global data centers—a figure that could range anywhere between a 35% and 128% increase.
But data centers run hot and use copious amounts of water to keep cool. According to a new preprint academic paper, global AI demand may account for 4.2-6.6 billion cubic meters of annual water withdrawal by 2027—that’s about half of the UK’s annual usage, according to the Financial Times.
Major technology companies are using more and more water just training their increasingly powerful generative AI models: Google and Microsoft self-reported annual water consumption increases of 20% and 34% respectively while training their models.
If AI continues on its current, staggering trajectory, it very well may push us closer to an energy crisis that’s impossible to ignore.