When President Donald Trump last week triumphantly announced Stargate, a massive data center expansion to fuel the energy demands of artificial intelligence (AI) in the U.S., project partners OpenAI CEO Sam Altman, SoftBank CEO Masayoshi Son, and Oracle Corp. founder Larry Ellison – who flanked the president – beamed.
“I think this will be the most important project of this era,” Altman boasted.
Son added, “This is the beginning of golden age,” echoing Trump’s proclamation that the U.S. would enter a “golden age” with him back in the White House.
The U.S., already the world leader in AI investment, could conceivably pump up to $500 billion over the next several years into what its backers claim is the “largest AI infrastructure project in history.” OpenAI went so far as to claim Stargate will create “hundreds of thousands of American jobs” and “generate massive economic benefit for the entire world.”
Then reality hit in a matter of days, starting with the skepticism of one Elon Musk.
“They don’t actually have the money,” Musk tweeted, while re-affirming his long-running feud with Altman.
Indeed, a Financial Times report said the project hasn’t secured funding yet, and won’t receive government financing. OpenAI and SoftBank will each commit more than $15 billion.
The litany of criticisms and concerns don’t end there. Industry experts have serious doubts about jobs creation, environmental impact, the project’s centralization of power, and the belief that extravagant spending will translate to AI dominance – especially in lieu of Chinese AI startup DeepSeek and its claims of advancements in reasoning models at a comparatively minimal cost.
Steven Dickens, CEO and principal analyst at HyperFRAME Research, believes the project’s biggest hurdle is fulfilling the promise of up to 100,000 jobs. “While the initial build-out of a datacenter is labor intensive, once built each datacenter employs very few workers. By design, datacenters are ‘lights out’ and the server per admin ratio has changed dramatically over the years with advances in automation and observability,” Dickens said in an email. “While I applaud this initiative for its audacious goals, and think it vital for national security, I highly doubt the jobs projections.”
Then there is the practical matter of whether there are enough high-end GPUs to meet the soaring scale of Stargate’s demand – and what impact so many large data centers would have on the environment.
Spending at $10 billion per year on Stargate translates to somewhere between 100,000 to 200,000 high-end GPUs per year, according to an IDC report. Yet in the last four quarters ending in Q3 2024, the entire server market consumed just under five million high-end discrete GPUs.
“Demand for GPU servers from Stargate may place a significant strain on GPU supply, which had just begun to ease during 2024 after a very supply-constrained 2023,” IDC concludes.
Even if there are enough GPUs to feed this massive data center expansion, there are concerns about their carbon footprint. An analysis conducted by The Guardian, based on data between 2020 and 2022, revealed actual emissions of data centers operated by Microsoft Corp., Alphabet Inc.’s Google, and Amazon.com Inc. were 662% higher than what has been officially reported. And that was before the AI stampede of the last few years which is only expected to intensify.
Such are the immediate concerns over Stargate. Equally troubling in the long haul is the concentration of AI development power among OpenAI, SoftBank and Oracle that Stargate represents.
“This centralization could limit competition, increase surveillance, and deepen societal inequalities by consolidating too much power,” Alan Orwick, co-founder of Quai Network, said in an email. “The opaque nature of these AI systems, often referred to as ‘black boxes,’ also raises concerns about accountability and inherent biases. While the technological advancements of Stargate are clear, its concentrated control may jeopardize ethical oversight and open the door for misuse of AI for profit or political agendas.”
A lack of robust oversight, argues 404 founder Ben James, could lead to surveillance, bias, and misuse of personal data while prioritizing profit over public interest.
“The project also risks escalating an international AI arms race, pressuring rival nations to prioritize speed over safety and ethics,” James warned. “To avoid becoming a dystopian cautionary tale, Stargate must strike a balance between bold ambition and strong ethical safeguards, ensuring societal benefit remains a priority.”
As for the fate of enormous data centers, when and if they are constructed?
“Are we creating the next ‘Rust Belt’? OK, so we build out these massive AI data centers with today’s tech, and then try to power them with massive new electric generation capabilities,” muses Jack Gold, principal analyst at J. Gold Associates. “But where will we be in five to 10 years? Will we still need these massive sites given how quickly technology moves? In the 1960s we built massive mainframe sites, and now we do more on our smartphones than the early mainframes did. I suspect in 10 years, we’ll be looking at these massive shells and wonder why we ever needed them, much like the ruins left behind in so many other manufacturing industries.
“Instead of the Rust Belt, are we creating the future equivalent Silicon Belt?” he asked.
