Michael Burry is certain: AI is a bubble. The American investor – known from the film The Big Short – made a fortune in 2008 by predicting the credit crisis. Now he is betting on a fall in the share price of AI companies. With his hedge fund Scion Capital, Burry recently bought millions of Nvidia and Palantir put options, which represent a value of approximately $1 billion.
Palantir CEO Alex Karp responded at news channel CNBC with an angry tirade when he heard about Burry’s news last week and his stock took a hit. “This is super strange. The two companies where he short it is precisely the companies that make all the money.” The stock prices of Palantir (which makes AI software for defense) and chip group Nvidia have recovered quickly from the short-lived dip in recent days.
But still, Burry is not the only one who wonders whether investments in AI are going too fast. There is little doubt that artificial intelligence will have a major impact on many sectors. The main question is: how will the hundreds of billions that Big Tech spend on data centers and chips be recouped? And can the rest of the world keep up with the current pace of construction?
Withdrawing investors
The British business newspaper Financial Times (FT) reported on Tuesday that investors are currently withdrawing from debt positions of American tech giants such as Alphabet, Oracle, Microsoft and Meta. in a way in which they run as little financial risk as possible.
A striking example of this is Hyperion, a data center yet to be builtof Meta in the American state of Louisiana, which should become the same size as Manhattan. To pay for the project, Meta entered into a partnership with an investment company, which finances 80 percent of the investment. Once Hyperion is ready, Meta will lease the new data center. If things go wrong with Hyperion, Blue Owl Capital in particular will be in trouble.
Management consultancy group Bain estimated last month that major tech companies will need to invest around $500 billion per year to realize all AI ambitions for 2030. This can only be done by taking out huge loans.
According to the FT private investors have realized in recent weeks that they are being used to finance this investment wave. Meanwhile, regular investors benefit from the persistently high profit margins and rising stock prices of tech companies. In short, the investors think it is high time that everyone started paying for the infrastructure of the future.
Also read
AI: the bubble that becomes too big to burst
Experiment phase
The big question is whether the technology will be used as fanatically as tech companies hope. Language model ChatGPT has been around for three years now, but last week research by consultancy firm McKinsey showed that the vast majority of companies
The majority of businesses still see artificial intelligence as a way to save costs and get rid of staff. Instead of what tech companies hope for:
And there are more hurdles to overcome. Real estate developers can hardly keep up with the requests from tech companies for new data centers. And data centers need so much power that, in the absence of a good power supply, tech companies build their own solar parks and enter into contracts with nuclear power plants.
In the meantime, we have to wait until the AI models start generating money and whether tech companies will succeed in raising money again and again until that moment comes. OpenAI, the company behind ChatGPT, would according to The Wall Street Journal suffered a record loss of $12 billion last quarter. Altman’s company is said to be working on plans for an IPO to continue financing these losses.
Because according to Altman, no one should doubt that AI is the future. “Whether we lose 500 million, 5 billion or 50 billion a year, I don’t care,” Altman said about last year the rate of cash burning of his business. “It will be expensive, but it will be worth it.”
NEW: Give this item as a gift
As an NRC subscriber you can subscribe every month 10 items give as a gift to someone without an NRC subscription. The recipient can read the article directly, without a paywall.