Big Tech’s plans can bankrupt the U.S. economy
Why are the big tech companies inflating a bubble they know will pop?
In the three years since ChatGPT’s release, excitement from the capitalist leaders of the U.S. has reached a fevered pitch. Trillions of dollars of investment have been thrown into designing and creating graphics cards, data centers, and AI models in a race towards a claimed new industrial revolution on the scale of what the internet delivered. At the same time, the lack of any real profitable investment has led to leaders of industry and finance such as OpenAI CEO Sam Altman and International Monetary Fund director Kristalina Georgieva to acknowledge that an economic bubble has formed.
This acknowledgment has not led to a slowdown in investment; in fact, it is accelerating. Business research firm Gartner estimates that worldwide AI investment for 2025 at $1.5 trillion, up from $1 trillion in 2024, and going even higher in 2026. Ruchir Sharma at The Financial Times put it best, “America has become one big bet on AI.”
So, what is a bubble?
In layman’s terms, it is when something is being sold more than what it is worth to the person buying it. A bubble arises at any time where labor and capital are invested into a speculative economic sector far above their ability to integrate back into the general exchange of goods that the rest of the economy is based on. Since the free market is based around speculation and not rational planning, investors are free to guess what consumers may want in the future and spend surplus wealth they have accumulated towards creating the supply for this imagined demand. If the demand is not being realized, and the investor runs out of money or must pay back debts that they cannot afford, the bubble “pops” and the investors must sell off what they own to repay debts they have taken on while workers employed in this sector are thrown out of work.
A famous example of this is the dot com bubble, in which companies piled on debt to build out fiber internet infrastructure, and startups raised investment money to run internet sites before enough people were ready to regularly use the internet at a rate to make this investment worthwhile. When it popped, internet infrastructure companies and web-based startups went bankrupt.
The existence of a bubble has nothing to do with whether a technology or speculative sector can be useful or one day could be integrated into the larger economy, just that it is not possible today. The dot com case makes this explicitly clear as the internet and internet services now constitute one of the only parts of the U.S. economy that is growing profits year over year.
The case that AI is a bubble is based on simple terms, no company that has made investments into AI tooling has produced any profit on these services. Historically, Silicon Valley has been happy to operate under this model by using their super-profits to subsidize new services and products sold at a loss until they capture a market and then begin raising prices. Strategies like this allowed Netflix, Uber, and Airbnb to take over existing markets in the U.S. The difference with AI as opposed to these previous companies that have successfully turned a profit is the scale of investment. AI investment is on the order of trillions of dollars now, with the leaders of these companies clamoring for more.
Sam Altman has even begun floating the idea of government support to continue building out AI infrastructure, while investors such as Softbank are now having to sell off all their other assets to keep making AI commitments.
OpenAI alone has $1.4 trillion in financial commitments to meet over the next eight years. Contrasting this with their estimated revenue of $12 billion for 2025, which amounts to less than 7% of their current financial commitments in a given year, there is no reason to believe that the debts they have will be paid. Many other companies have similarly outlandish balance sheets when it comes to costs vs revenue generated by AI investments.
Given that we are looking at a bubble, the question then becomes how does it pop? It is hard to say definitively what will trigger the crisis, but there are few glaring risk points to look at. First is Nvidia, which is the chip manufacturer whose designs are the entire material basis for why AI is valued as much as it is. The story they have sold investors is that AI is going to transform the global economy using their graphics cards; therefore they will keep selling more year after year, and therefore an investment in them is an investment in the future of the AI revolution. If they manage to keep selling more chips, that story sounds good to investors. If they ever have a miss on sales through an unexpected slowdown, it completely collapses.
Signs of this story being false are already appearing through two examples. One is the rising amount of graphics cards that are sitting in Nvidia warehouses. MarketWatch reports that the amount of inventory being kept at Nvidia has doubled in value since last year, meaning that while Nvidia is still increasing sales, it is also over-producing cards, a classic sign of over-estimating demand. The other is vendor financing of companies like CoreWeave, and OpenAI, which are companies who are buying Nvidia chips and in return are getting investment from Nvidia, creating a circular flow of capital that doesn’t actually connect back to general trade.
The vendor financing points to the second risk of the bubble pop, which are the companies who are at the forefront of the AI build-out and took on loads of debt and financial obligations to meet it. OpenAI is the most obvious case, a firm that has never reported a profit that has secured billions of dollars in capital and promised over a trillion dollars in financing to the rest of the tech industry.
While tech companies have an interest for now in not bankrupting OpenAI, as its popularity is boosting their stock evaluations, all it takes is for one or two companies to demand their money and the house of cards crumbles. A less heard of but similar case is that of CoreWeave, which is a data center company that took on immense amount of financial debt to build out AI data centers using GPUs from Nvidia and is now facing a crisis as it cannot find firms to lease them to. Both OpenAI and CoreWeave are being bailed out by the super-profits from Nvidia, which is investing back into these companies to prop them up, creating an uroboros of capital. This can only sustain itself so long as Nvidia is able and willing to feed the companies that have yet to convince the rest of the economy that they are needed.
Lastly, there is the electrical ceiling that looms over the entire data center build-out to support AI. Data centers already had high energy costs, but AI data centers are orders of magnitude more power hungry. While the U.S. has a capital and labor supply that can secure massive amounts of AI GPUs and data centers, there are now serious constraints on how much power is available to build the centers. Nvidia’s modest AI centers in Santa Clara, in the center of Silicon Valley, can’t be powered on due to lack of energy capacity; Amazon is currently suing an Oregon power supplier due to failed commitments to meet power demand; and local communities are now beginning to vote down AI data centers being built near them due to the impact they will have on their electricity bills.
In response, tech companies are trying to invest in nuclear energy and their own on-site power sources—but it will be too little, too late. The Energy Information Administration estimates that construction of a nuclear power plant takes five or more years, and if it is a new design, design approval alone can take up to five years. Power plants take several years to start up and operate, and the AI financial obligations that companies such as OpenAI and CoreWeave must meet are happening right now. This has led to more data centers turning to on-site power generation, as opposed to connecting to a grid, which will further spike operating costs.
A crisis for you, an opportunity for Big Tech
The big financial risks here beg the question, why would tech companies acknowledge a bubble and continue to inflate it? The answer is that for the biggest tech companies, triggering a recession can have positive outcomes. Big Tech has existing profitable businesses that the AI bubble does not threaten, and wiping out smaller tech companies who are betting on AI can be beneficial. For one, it destroys competition from startups seeking to unseat the entrenched leaders of industry. For two, it would also spike unemployment for tech workers, who have consistently commanded high wages compared to the average worker in the U.S. Higher unemployment would allow tech companies to discipline workers in the tech sector further and drive down their wages, increasing their profits.
Lastly, if the bubble popping becomes an existential threat to the biggest players in the tech industry, they can always run to the White House for a bailout. The Trump admin has already done so for Intel by doling out $5.7 billion in return for a 10% stake in the company. In the final analysis, while an AI bubble burst will have huge negative consequences for the working class in the U.S, as well as sectors of the middle-class and even capitalists, Big Tech stands to lose far less than their start up competitors. Even if they end up in hot water financially, they can always bail themselves out with the coffers of the US government by leveraging their direct line to the White House.
For workers in the U.S. who are not directly employed in the AI industry or one of its key inputs, the recession is already here. Unemployment is at a four-year high, foreclosure rates are up 19% from last year, and car repossession rates are on par with the Great Recession. The Trump administration is staving off news of a recession by deliberately interfering with and delaying key objective economic indicators such as the monthly BLS and CPI reports that tell banks and investors employment and inflation data. The stock market’s boom under year one of Trump gives cover for the capitalist class to claim all is well, but the market shocks that occurred during liberation day in April as well as previous stock market corrections demonstrate that this narrative can collapse within a single day.
A crash in the economy will surely mean more pain for workers in the U.S. and abroad. Trump’s threats to Venezuela are providing a preview of what is to come when the AI industry collapses. Once the market corrects itself and capitalists realize they can’t find anywhere to invest domestically, the push to find new markets abroad will intensify. The recent Trump foreign policy document makes this very clear. The “Trump Corollary” to the Monroe Doctrine will be enforced on Latin America to discipline left-wing governments and force U.S. investment onto countries that cannot stop US intervention.
For nationalization and workers’ control
Working people in and outside of the United States need a different path. The government could simply seize the existing Big Tech institutions and nationalize them under workers’ control. Workers at these firms would form unions to represent them, in addition to democratically elected committees to manage the job sites.
In the immediate period, this course would wrestle control of the tech industry away from the big capitalists, whose iron grip has been dictating not only the direction of the U.S. economy but large segments of the world economy as well. In the longer view, it would allow for a democratic and rationale planning of production. For example, AI research and development could be capped to what workers think is a reasonable amount of money to spend researching it, and the applications that are researched could be determined on a democratic basis as well to prioritize uses that benefit people as opposed to helping generate content meant to automate other’s labor or to surveil people.
Doing this would actually accelerate the development of useful AI technologies as it would avoid the scramble to chase after sci-fi dreams of an artificial general intelligence that even capitalists are admitting won’t happen even as they pour trillions of dollars into the venture.
The Great Recession of 2008 is still in living memory for most adults; the fallout from it has scarred generations. One generation of the U.S. working class had their job opportunities evaporate before their eyes, and another had their homes snatched away while they tried and scrape the funds together to try again on home ownership in several years. The paltry reforms that came out of that crisis were repealed under the first Trump administration. Less than twenty years later, the AI bubble has the potential to have far graver consequences as it threatens the entire U.S. economic system. This pattern of speculation, crash, and recovery is one that is doomed to repeat itself as long as the economy is structured around benefiting an ever-smaller group of titans of industry and finance. Prying the institutions of Big Tech away from the elites who own and run it is a necessity not just for the millions who toil under their regime, but for the good of working people across the U.S. and abroad.
First published here by Workers’ Voice
