The lure of AI to transform the world has led the stock market to new heights and revitalized the venture capital community after the 2022 downturn. But as stocks like Nvidia — the darling of generative AI stocks with its chip monopoly — recently retreated from all-time highs, a few voices have spoken up about what they view as the false promises of the new technology.
Goldman Sachs is the most prominent of these critics, who also include MIT professor Daron Acemoglu and Silicon Valley pioneer Roger McNamee, a cofounder of Silver Lake Partners. Even VC heavyweight Sequoia Capital has raised concerns.
The main question for Goldman’s Jim Covello, head of global equity research, is whether the $1 trillion likely to be spent on AI in the next few years will earn an appropriate return on investment.
“What trillion-dollar problem will Al solve?” he asked, noting that “replacing low-wage jobs with tremendously costly technology is basically the polar opposite of the prior technology transitions I’ve witnessed in my thirty years of closely following the tech industry.”
To justify its extraordinarily high costs, AI “must be able to solve complex problems, which it isn’t designed to do,” he explained in a recent Goldman report on the topic.
The technology is so expensive that it won’t even cut costs by replacing humans with machine learning. “We’ve found that AI can update historical data in our company models more quickly than doing so manually, but at six times the cost,” he said. Costs would have to come down dramatically to make automating tasks with AI affordable, he added.
The argument of its proponents is that AI is in its infancy, much like the internet of the 1990s dot-com boom, and that costs will eventually come down. But even then, Covello noted, the internet had a cost advantage. “Amazon could sell books at a lower cost than Barnes & Noble because it didn’t have to maintain costly brick-and-mortar locations.
The idea that technology typically starts out expensive before becoming cheaper is revisionist history,” he said.
It isn’t just the high cost that Covello worries about. He simply doesn’t think that AI will be the breakthrough technological invention people are expecting. So far, there is no “killer app” for AI, as even his more bullish Goldman colleagues acknowledged in the report.
Covello’s views largely dovetail with those of MIT professor Acemoglu, who was interviewed by Goldman in its report, where he argued that the productivity gains from AI are likely to be minimal.
Acemoglu — who also wrote a paper criticizing AI several months ago — believes that AI will affect less than 5 percent of all tasks in the next ten years, because only a quarter of those that could be done by AI will be cost-effective. He too disagrees that the technology will become less expensive over time and argues that AI model advances likely won’t wow the world.
“The largest impacts of the technology in the coming years will most likely revolve around pure mental tasks, which are non-trivial in number and size but not huge, either,” he told Goldman.
“I question whether AI technology can achieve superintelligence over even longer horizons because it is very difficult to imagine that an LLM [large language model] will have the same cognitive capabilities as humans to pose questions, develop solutions, then test those solutions and adopt them to new circumstances,” he added.
As a result of these concerns, Acemoglu forecasts AI will increase U.S. productivity by only 0.5 percent and GDP growth by only 0.9 percent cumulatively over the next decade.
McNamee, who lived through the dot-com bubble (and bust) and now heads VC firm Elevation Partners, is also waving red flags. On X, he laid out a number of issues that he says Goldman Sachs didn’t even get to.
“Goldman is right on, but does not address other crippling issues facing AI,” he posted on X last week. He also pointed to a Sequoia Capital report that estimated the “industry needs $600 million in revenue to justify the existing investment in compute and cloud.”
The “crippling issues” McNamee addressed include AI’s incredible thirst for both power and water. “Generative AI is breaking the power grid nationally and accelerating climate change,” he said. As for water usage, he noted that “one report suggests 1/2 liter of water is ruined every time you query a chatbot.”
Other glaring problems revolve around copyright, privacy, national security, disinformation, and the like, McNamee posted.
“LLMs are not intelligent,” he cautioned. “They use statistics to find the most suitable next word, paragraph, image. They only know their training set. What they do best is to BS you. More evidence supports the view that LLMs are a scam than the Next Big Thing. Beware.”
That said, few expect the AI arms race to end anytime soon. “This is not the first time a tech hype cycle has resulted in spending on technologies that don’t pan out in the end; virtual reality, the metaverse, and blockchain are prime examples of technologies that saw substantial spend but have few — if any — real world applications today,” said Covello, who like McNamee was around during the last tech mania cycle. (He was Goldman’s star semiconductor analyst.)
“Overbuilding things the world doesn’t have use for, or is not ready for, typically ends badly,” the analyst said, noting that “the NASDAQ declined around 70 percent between the highs of the dot-com boom and the founding of Uber.”
He argued that if important use cases aren’t discovered in the next 18 months, “investor enthusiasm may begin to fade” though he thinks “sustained corporate profitability will allow sustained experimentation with negative ROI projects.”
So far, betting against AI has been costly. As Nvidia became the most valuable company in the world, with a market cap of more than $3 trillion, it also became the least profitable short in the second quarter. On a mark-to-market basis, shorts lost more than $9 billion betting on Nvidia, according to S3 Partners.
Said Covello: “One of the most important lessons I’ve learned over the past three decades is that bubbles can take a long time to burst.”
