In the early weeks of October, hardware development company Nvidia announced its plan to invest 100 billion dollars in OpenAI, the software company responsible for the A.I. language model ChatGPT, among other significant generative A.I. programs. In turn, OpenAI has announced that they will be engaging in a partnership with 10 gigawatts of new AI data centers and other infrastructure built with Nvidia technology. Despite the new infrastructure not being deployed until 2026, this new partnership has caused Nvidia’s value to skyrocket. The hardware company has been growing rapidly in the past few years, and more so now with A.I. company partnerships; however, market experts have expressed worries that this ouroboros of companies investing into one another without any clear output for consumers, combined with significant amounts of debt in the U.S. economy, is rapidly blowing up a bubble that could pop at any time.
A particularly alarming fact is that OpenAI has been losing three times more money than it makes. This is not an exaggeration. According to OpenAI themselves, in the first half of 2025, they earned $4.3 billion against a loss of $13.5 billion. Another A.I. company, Anthropic, which makes Claude, lost $5.3 billion last year. And what’s more, an MIT study found that 95% of companies which try to invest into A.I. see no return.
The A.I. bubble is not sustainable, not just because of abstract stock market forces and massive losses, but because there is simply not enough electricity to power all of the A.I. data centers that are rapidly being built. A.I. uses an almost unbelievable amount of energy. Researchers expect A.I. data centers to use up to 326 terawatt-hours per year by 2028, equivalent to the energy needs of 31 million homes. The Three Mile Island Nuclear Power Plant is set to restart electricity production in 2027. However, none of the energy will be used to provide power to the houses and businesses in the area. Instead, Microsoft will use all of the output from an entire nuclear power plant to provide power for just one of its data centers. In “Data Center Valley,” an area of Virginia with hundreds of A.I. data centers, residents faced blackouts caused by too much electricity being used by these massive data centers. The growing electricity demand caused by A.I. is also slowing down America’s transition to green energy, with more than 60% of data centers powered by fossil fuels.
Do the use cases of A.I. success outweigh the massive costs? The most obvious use case is in academics, as many of our readers surely know. But is there anything beyond that? In fact, OpenAI itself released a chart showing that usage of ChatGPT plummets by nearly 50% in the month of June, when schools let out. If a solid 50% of usage comes just from kids needing homework help, that doesn’t bode well. And there’s even less use cases for Sora and other video-generation A.I. It’s hard to think of a use case for A.I. video outside of lazy political propaganda or TikTok content farms—and audiences will probably get sick of these eventually. Is that worth the massive losses OpenAI takes to generate “AI slop?”
In The Hoot’s recent interview with President Arthur Levine, AI was a recurring theme. President Levine seems convinced that A.I. is the future of higher education and Brandeis’ ticket to the top of university rankings. With a series of A.I.-centered projects on the docket for the next few years, Levine and the campus administration as a whole seem to disregard any data on the negatives of A.I. usage. In the interview, a member of our editorial team pressed Levine on the news stories of A.I. being linked to multiple suicides and negative environmental impacts caused by generative A.I. usage. Levine dismissed these possibilities, calling the possible negative impacts of A.I. necessary to the improvement of technology and conflating being anti-A.I. with being anti-technology as a whole.
If—or when—the bubble pops, what happens next? Generative A.I. will probably continue to exist, and perhaps even improve. If this follows the trend of the dot-com bubble, the days of every company rushing to integrate A.I. into their workflows at all costs will probably come to an end, as will the widely held belief that A.I. will replace human creativity in every capacity. Before the dot-com bubble burst in 2000, many were convinced that online retail would destroy brick-and-mortar stores. While some shopping malls have closed, online retail clearly hasn’t replaced in-person shopping trips, while once-hyped companies such as Pets.com and WorldCom have crashed into obsolescence. Could a similar future be in store for today’s A.I. companies? On a student level, although there is clear opposition to the use of A.I. among the student population, many of us still incorporate the technology into our study planning, and it even appears in our day-to-day life without our knowledge, with Google Gemini’s A.I. answers as a primary example. While the future of A.I. is uncertain and out of our control, we can at least consciously decide when and if we participate in the A.I. economy and how it shapes our individual lives.
