The Big AI Con
This week, technology investor extraordinaire, Roger McNamee, gave an interview on CNBC that is a must watch for anyone exposed to the broad U.S. stock market. In it, he lays out why he believes the current AI bubble is nothing more than a “big con”:
I do think that what these companies have done, and I think Altman at OpenAI has been a master of this, is that they have spun a tale and they have gotten politicians, CEOs, journalists, and investors to buy into this thing without spending a lot of time thinking about whether it’s going to work. Everybody should go back and watch the classic film “The Sting” with Paul Newman. You know, you just need to remind yourself how a really big con works. And I don’t think there’s ever been a more successful pitch of fantasy to a more strategic group of, uh you know, marks than you’ve seen with this technology. And I really tip my hat to these people. I mean tech’s been exaggerating what it can do for decades. But it was always by 5, 10, 15%. This notion that you would promise something that you couldn’t possibly deliver and that everybody would buy that even as the evidence came in that it doesn’t work that well that that part is really amazing to me and you know I don’t know how this thing shakes out.
For those that don’t know, Roger is one of the most successful investors of his generation, buying into both Facebook and Google in the very early days. Because his views regarding the companies, now Meta and Alphabet, have done a complete 180, investors ought to try to understand the reasoning behind the change of heart.
As to the contention that current AI products don’t “work that well,” a new study was released this week verifying that point. In a study of software developers using AI, asked how much they believed the products would reduce their time spent they estimated an average of 24%. In reality, “Current AI tools actually slowed down task completion time by 19%.”
Software coding has been perhaps the biggest success story of the generative AI boom, so far. So the fact that it is, in truth, damaging productivity rather than improving says a lot about the hopes for AI playing catalyst for a productivity boom throughout the economy.
The trouble, as McNamee point out, is that these companies have already spent hundreds of billions of dollars developing these products that don’t “work that well.” Nowhere is this more obvious than at Meta, where Mark Zuckerberg has gone into panic mode and starting hiring AI engineers with pay packages running into the hundreds of millions of dollars.
“The poaching of engineers feels like rearranging deck chairs on the Titanic,” writes McNamee in regard to these moves. And, listening to the engineers at Meta themselves, it sounds about right. “You’ll be hard pressed to find someone that really believes in our AI mission. To most, it’s not even clear what our mission is,” wrote one in a letter made public this week.
Of course building AI products that underwhelm your customers and appear to have few viable uses would be discouraging. But to hear it from the horses mouth is something else. And the fact that Zuckerberg just throws more money at the problem is not likely to help.
It certainly isn’t helping OpenAI, one of the companies losing engineers to Meta’s poaching. The company already spends more than 100% of its revenues on stock-based compensation, a figure that is not included in its estimated $10 billion operating loss this year. Ramping this up to retain engineers only exacerbates what is already a serious problem.
Meanwhile, DeepSeek is apparently taking marketshare from OpenAI and its peers by offering, “the same quality but 17 times cheaper,” as one customer put it. So there is pressure on both sides of the income statement: revenue growth is threatened by low-cost Chinese products while expenses are soaring due to the cost of building out data centers and hiring and retaining engineers.
As Jim Chanos points out, this makes for a pretty precarious setup: “There is an ecosystem around the AI boom that is considerable as there was for TMT back in ‘99 and 2000. But it is a riskier revenue stream because if people pull back, they can pull back CapEx very easily and that immediately shows up in disappointing revenues.”
This may be why Nvidia insiders (and hedge funds) have taken to the open market once again to sell shares hand over fist. They likely understand as well as McNamee does that, as he puts it, while don’t know exactly how it shakes out, there’s a very good chance that there’s a, “Train wreck coming.”
And because the stock market has never been more concentrated in the tech sector than it is today, this should be of great concern to investors of all kinds.
Literal Ponzi Finance
Hyman Minsky wrote about the economic and market cycle by dividing it into three phases ending with the Ponzi phase. While the cycle begins very conservatively, in this final phase, borrowers not only fail to make principal payments out of cash flow, they cannot even make payments on interest expense. In other words, constant new borrowing is needed simply to service existing borrowing.
There are all sorts of signs that we are now in the Ponzi phase of the cycle. First and most obvious may be the proliferation of payment-in-kind lending. This entails a struggling borrower, rather than making interest payments, simply adding those liabilities to the principal balance. And it has allowed many companies financed by private credit to avoid default in recent years.
Another example can be seen in the fact that the share of companies within the public markets that currently qualify as Ponzi financing schemes, and are unable to service their debt out of cash flow, is at a record. In fact, the vast majority of small cap companies fall in this category today.
But I’m sure that not even Minksy ever imagined public companies brazenly operating literal Ponzi schemes as so many are doing today. Case in point is Strategy (formerly Microstrategy). The company has pioneered the “Bitcoin treasury” model, in which the primary business model takes a back seat to the accumulation of cryptocurrency made possible by sales of new securities.
Strategy’s most recent security sales, though, represent an important change in the pattern. Rather than using the proceeds to buy more crypto, the company is, “effectively reserving the right to use money from new securities to prop up old ones, reassuring investors that it can keep issuing fresh paper to cover dividend payments. This in turn is meant to give them confidence to buy the future rounds of preferred shares,” as the Financial Times reports.
This not only meets the definition of Minsky’s Ponzi finance phase of the cycle, it also meets the definition of a pure Ponzi scheme. Using new money coming in to pay previous investors is exactly what Bernie Madoff, and every other Ponzi scheme operator in history, did.
What is astounding about all of this is that, not only is there no pushback at all, Wall Street is in fact racing to mimic the Strategy model as fast as it can. Companies are converting to the Bitcoin treasury model left and right in a, “multibillion-dollar capital markets experiment,” aimed at cashing in on investor demand for the most speculative of financial endeavors.
As Charlie Munger said, “Wall Street will sell shit as long as shit can be sold.” But the fact that there is so much demand for what amounts to outright Ponzi schemes says much about the precariousness of the current environment.
In his Financial Instability Hypothesis, Minsky famously warned, “If an economy with a sizeable body of speculative financial units is in an inflationary state, and the authorities attempt to exorcise inflation by monetary constraint, then speculative units will become Ponzi units and the net worth of previously Ponzi units will quickly evaporate.”
Certainly, the economy today is in an inflationary state like we haven’t seen for quite some time. The central bank has been attempting to exorcise it without success. In fact, for a number of reasons inflation could be poised to surge again, making the challenge that much more difficult for Jay Powell & Co.
And the rapid growth in “Ponzi units” that we have seen within the markets is likely due to these precise dynamics. Minsky continues, “Consequently, units with cash flow shortfalls will be forced to try to make position by selling out position. This is likely to lead to a collapse of asset values.”
This is the stage that we have yet to see. But, as we learned during the GFC, once the Ponzi stage of the cycle has gone on for so long, a painful reckoning becomes inevitable. And at that point, it won’t just be the traditional Ponzi units that become a problem for markets; the literal Ponzi schemes operating today are likely to lead the, “collapse in asset values.”
