DIGITAL LIFE
Billions are flowing into the artificial intelligence sector, boosting giants like Nvidia and OpenAI and even sustaining the growth of the US economy. But a new report from Deutsche Bank suggests that the "parabolic" pace of investment is unsustainable—and the bubble may soon burst.
In recent years, artificial intelligence has become the engine of technological optimism and, to some extent, of the US economy itself. From mega-investments in chips to data center projects that exceed the GDP of entire countries, the money seems endless. But, according to Deutsche Bank, there are clear signs that the euphoria may be masking profound risks.
In a note signed by George Saravelos, the bank's strategist, the warning is blunt: "AI machines are literally saving the US economy right now." He claims that, without technology-related spending, the country would be very close to a recession in 2024.
This is consistent with data already published by the Wall Street Journal, which showed that investments in AI infrastructure contributed more to US GDP growth this year than all household consumption combined.
Among all companies, Nvidia stands out as the main driver of the expansion cycle. For Saravelos, the chipmaker is "bearing the brunt of US economic growth."
The problem? This cycle can only continue if the company's growth is exponential—something difficult to sustain for long. "For technology to continue driving GDP, investment needs to remain parabolic. This is highly unlikely," the analyst wrote.
Extreme market concentration...The risk isn't limited to Nvidia or the chip sector. For Torsten Sløk, chief economist at asset management firm Apollo, there's a structural problem: the level of concentration in the S&P 500 is "extreme," leaving investors dangerously exposed to the AI sector.
In other words, a slowdown in the artificial intelligence hype could have ripple effects across the stock market.
Besides the financial risk, there's the physical cost of sustaining the AI revolution. According to Bain & Company, the computational demand for artificial intelligence is growing at more than double the rate predicted by Moore's Law.
If this trend continues, by 2030 the world will need to spend $2 trillion per year just to meet AI's computational demand. The problem: there's still about $800 billion needed in infrastructure to keep up with this pace.
The billion-dollar investment, so far, hasn't been reflected in a proportional return. An MIT study showed that only 5% of companies that adopted generative AI tools saw significant revenue acceleration. The rest saw modest or no gains.
The disparity between expectations and results reinforces the perception that the market may be overvalued and that the much-promised "transformation" of AI has not yet translated into tangible profits.
What lies ahead...The central question, according to analysts, is whether the sector will be able to justify the colossal investments before confidence ebbs. While some argue that the learning and adoption curve is still in its infancy, others believe the bubble bursting is only a matter of time.
In the end, as the report ironically states, perhaps all that's needed is a "small" investment of US$500 billion for everything to finally make sense. Who's up for it?
mundophone
No comments:
Post a Comment