Will the race to artificial general intelligence (AGI) lead us to economic prosperity or end in a 2008-style collapse? The answer costs trillions of dollars.
This number is staggering. An estimated $2.9 trillion (£2.2 trillion) is being spent on data centers, the central nervous system of AI tools. Nvidia, a company that makes chips that power cutting-edge AI systems, has a market capitalization of more than $4 trillion. and the $100 million sign-on bonus that Mark Zuckerberg’s Meta gave to top engineers at OpenAI, the company behind ChatGPT.
All of these staggering numbers are backed by investors looking to make trillions of dollars in profits. AGI, a theoretical state of AI in which systems can gain human-level intelligence across a variety of tasks and replace humans in white-collar jobs such as accounting and law, is the keystone of this economic promise.
This offers the prospect that computer systems can perform profitable work without the costs associated with labor costs. This is a very win-win scenario for the companies developing the technology and the customers deploying it.
If AI companies fail, there will be consequences. The U.S. stock market, which has risen sharply due to the performance of tech stocks, will fall, potentially damaging people’s personal finances. The bond market, which has been engulfed by the data center boom, could receive a shock that spreads elsewhere. GDP growth in the United States, which has benefited from AI infrastructure, could slow, which would have ripple effects on the interconnected economy.
David Kahn, a partner at Sequoia Capital, a major Silicon Valley investment firm, says technology companies now need to realize AGI.
“There is not enough outside of AGI to justify any proposed investments over the next 10 years,” he said in a blog post published in October.
This means there are many challenges to advancing advanced AI, and trillions of dollars are being poured into infrastructure and research and development to achieve it. Joshua Bengio, one of the “godfathers” of modern AI, says progress in AGI could stall, which would be bad for investors.
“It is clear that we may hit a wall. There are unforeseen difficulties at the moment and solutions will not be found soon,” he says. “And that may be true [financial] crash. Many of those currently spending trillions of dollars on AI also expect progress to continue fairly regularly at the current pace. ”
But Bengio, a prominent voice on the safety implications of AGI, makes it clear that continued progress toward a highly advanced state of AI is likely the end goal.
“The less likely scenario is that progress stagnates. The more likely scenario is that we continue to move forward,” he says.
The pessimistic view is that investors are supporting an unrealistic outcome: AGI will not occur without further progress.
David Bader, director of the New Jersey Institute of Technology’s Data Science Institute, said trillions of dollars are being spent on scaling up (a technical term meaning to grow something quickly) the underlying technology for chatbots known as Transformers, in hopes that building more data centers will be enough to increase the amount of computing power behind current AI systems.
“If AGI requires a fundamentally different approach, perhaps something we haven’t thought of yet, then we’re optimizing the architecture, and no matter how big we make it, we’re not going to get there. It’s like trying to get to the moon by building a higher ladder,” he says.
Nevertheless, major US technology companies such as Google’s parent company Alphabet, Amazon and Microsoft are moving forward with their data center plans, with the financial cushion of being able to fund AGI’s ambitions through cash generated from their hugely profitable day-to-day operations. This will at least give you some protection if the wall drawn by Bengio and Vader comes into view.
But there is an even more worrying aspect to this boom. Analysts at US investment bank Morgan Stanley estimate that $2.9 trillion will be spent on data centers between now and 2028, half of which will come from cash flow from “hyperscalers” such as Alphabet and Microsoft.
The rest will have to come from alternative sources, such as private credit, part of the shadow banking sector that has raised alarms at the Bank of England and elsewhere. Meta, the owner of Facebook and Instagram, borrowed $29 billion from private credit markets to fund a data center in Louisiana.
According to investment bank JP Morgan, AI-related sectors account for about 15% of investment-grade bonds in the U.S., even more than the banking sector.
Oracle, which signed a $300 billion data center deal with OpenAI, is increasing the amount of credit default swaps, a type of insurance against corporate default. High-yield, or “junk bonds,” which represent the riskier end of the debt market, are also emerging in the AI sector through data center operators CoreWeave and TeraWulf. Growth is also being financed by asset-backed securities (a form of debt backed by assets such as loans or credit card debt, but in this case the rent paid by tech companies to data center owners), which have proliferated in recent years.
No wonder JPMorgan says the AI infrastructure boom will require contributions from all sectors of the credit market.
“If AGI does not materialize as expected, there could be simultaneous ripples across multiple fixed income markets, including investment grade bonds, high-yield junk bonds, private credit, and securitized products, all of which are being used to fund this buildup,” Bader said.
Stock prices related to AI and technology also play a large role in the US stock market. The so-called “grand seven” of U.S. tech stocks (Alphabet, Amazon, Apple, Tesla, Meta, Microsoft, and Nvidia) account for more than a third of the value of the S&P 500, the largest U.S. stock market index, up from 20% at the beginning of the decade.
In October, the Bank of England warned of a “risk of sharp correction” in US and UK markets due to dizzying valuations of AI-related technology companies. Central bankers are concerned that stock markets could fall if AI fails to reach the transformative heights that investors expect. At the same time, the International Monetary Fund said valuations were heading toward dot-com bubble levels.
Even technology executives whose companies are benefiting from the boom recognize the speculative nature of the frenzy. In November, Alphabet CEO Sundar Pichai said there was an “irrational element” to the boom and that “no company will be immune” if it bursts, while Amazon founder Jeff Bezos said the AI industry was in “kind of an industrial bubble,” and OpenAI CEO Sam Altman said, “I think there are a lot of parts of AI that are kind of a bubble right now.”
To be clear, all three are AI optimists and expect the technology to continue to improve and benefit society.
But as Pichai acknowledges, when the numbers get this high, there is a clear risk of the bubble bursting. Pension funds and people who invest in the stock market will be affected by the collapse in stock prices, and the bond market will also be hit. There is also a series of “cyclical” transactions, such as OpenAI paying Nvidia in cash for the chips and Nvidia investing in OpenAI as a non-controlling stake. If these deals fall apart because AI implementation is delayed or hits a wall, things could become chaotic.
Some optimists argue that generative AI, a collective term for tools such as chatbots and video generators, will transform entire industries and justify the spending. Technology analyst Benedict Evans said the spending is not outrageous compared to other industries, such as oil and gas extraction, which operates at $600 billion a year.
“These AI capex numbers are large, but not impossible,” he says.
Evans added, “You don’t have to believe in AGI to believe that generative AI is a big thing. And most of what’s happening here isn’t, ‘Wow, they’re going to create God.’ It’s, ‘This is going to completely change how advertising, search, software, social networks, and everything else that is the foundation of our business works.’ That’s going to be a huge opportunity.”
Nevertheless, there are trillions of dollars in hopes for achieving AGI. For many experts, the consequences of getting there are alarming. The cost of not getting there can be significant.
