AI companies are looking to spend trillions of dollars on data centers to power their increasingly resource-intensive AI models – a huge sum that could put entire economies at risk if the bet doesn’t pay off.
As the race to spend more and more money on AI infrastructure grows, companies have become desperate to maintain cash flow. Companies like OpenAI, Anthropic and Oracle are draining existing lending markets – including junk loans, private loans and asset-backed loans – in increasingly desperate moves, as bloomberg reportsWhich is increasing concern among investors.
“These are numbers like none of us who have been in this business for 25 years have seen,” said Matt McQueen, managing head of global credit at Bank of America. bloomberg. “To make this work you have to cross paths.”
According to the publication, AI companies have accrued at least $200 billion in debt. A more realistic figure is likely to be significantly higher, as this estimate does not count undisclosed private deals.
Oracle announced over the weekend It is raising a massive $45 billion to $50 billion in debt and equity sales to create additional cloud infrastructure capacity, a plan that has once again highlighted persistent concerns over a growing AI bubble. The company’s efforts to build an AI data center have pushed the company firmly into negative cash flow, which could cost it billions of dollars in losses in the coming years.
Elon Musk’s plan Merged his space company SpaceX with xAI Eyebrows are also being raised ahead of a rumored blockbuster IPO that suggests the billionaire’s budding AI startup is looking to secure even more funding for highly ambitious plans including sending a data center into space.
Despite rising capital expenditure, the industry still has a lot of work to do to justify its heavy borrowings. Many AI companies have essentially given up even pretending that their short- or medium-term goal is to make money as they have become measured by “ambition, not success.” techcrunch AI Editor Russell Brandom Explained In a recent excerpt.
The technology is also starting to show diminishing returns with the release of each new model. Even the most powerful AI models are still struggling with the basics, while suffering from the same shortcomings that have plagued them for years, including persistent hallucinations.
Demand may also decline, making that entire loan even more difficult to justify. Early data suggests subscriber growth for online services like OpenAI’s ChatGPT may already be slowing. Meanwhile, OpenAI has already begun flooding its services with ads in a desperate bid to stop the bleeding, a move that CEO Sam Altman described as “last resort”As late as 2024.
The short-term forecast is starting to look grim. The growing mountain of debt could significantly increase borrowing costs, making AI data centers an even more expensive endeavor for already cash-strapped AI companies.
At least for now, investors are still looking for dollar signs – though many are also concerned that it’s only a matter of time before the bubble bursts.
“There’s this idea that if you can build a data center, there’s so much demand for data centers that you can’t lose it — it’s like selling beer to sailors,” explained Andrew Kleiman, co-head of private fixed income at SLC Management. bloomberg. “But whenever there’s a really innovative technology, there’s usually massive investment, and then improvements.”
More on rising debt: Major AI companies aren’t even pretending to make money
