A reply to Grantham’s AI warning

Well-known value investor Jeremy Grantham recently penned an essay titled, “The Great Paradox of the U.S. Market”, in which he warned, “Prices reflect near perfection yet today’s world is particularly imperfect and dangerous”.


In particular, he sounded the alarm over the bubble in AI stocks and cited the Gartner Hype Cycle as the main reason for caution:

But every technological revolution like this – going back from the internet to telephones, railroads, or canals – has been accompanied by early massive hype and a stock market bubble as investors focus on the ultimate possibilities of the technology, pricing most of the very long-term potential immediately into current market prices. And many such revolutions are in the end often as transformative as those early investors could see and sometimes even more so – but only after a substantial period of disappointment during which the initial bubble bursts. Thus, as the most remarkable example of the tech bubble, Amazon led the speculative market, rising 21 times from the beginning of 1998 to its 1999 peak, only to decline by an almost inconceivable 92% from 2000 to 2002, before inheriting half the retail world!



As much as I respect Grantham’s investment insights, he suffers from the value investor problem of being too early and overly reliant on valuation for his views. I reiterate my view that it’s still early in the bull cycle for AI stocks (see The Path to Magnificent Exuberance). Here’s why.



AI doubts

Where are we in the Gartner Hype Cycle? My best guess is the market is still ascending the curve, but it’s nowhere near the “peak of inflated expectations”. There is no question that the effects of AI are very real and they will change the way we do things over the next 10 years, but some questions are starting to appear about the hype.

The Economist reported that the rate of generative AI adoption and productivity gains may be slower than hoped and most AI use amounts to “window dressing”.
Some corporate giants are frantically experimenting to see what works and what doesn’t. They are hiring AI experts by the thousand…


This “use-case sprawl”, as one consultant calls it, can be divided into three big categories: window-dressing, tools for workers with low to middling skills, and those for a firm’s most valuable employees. Of these, window-dressing is by far the most common. Many firms are rebranding run-of-the-mill digitisation efforts as “gen AI programmes” to sound more sophisticated, says Kristina McElheran of the University of Toronto. Presto, a purveyor of restaurant tech, introduced a gen-AI assistant to take orders at drive-throughs. But fully 70% of such orders require a human to help. Spotify, a music-streaming firm, has rolled out an AI disc-jockey which selects songs and provides inane banter. Recently Instacart, a grocery-delivery company, removed a tool that generated photos of vendors’ food, after the AI showed customers unappetising pictures. Big tech firms, too, are incorporating their own AI breakthroughs into their consumer-facing offerings. Amazon is launching Rufus, an AI-powered shopping assistant that no shopper really asked for. Google has added AI to Maps, making the product more “immersive”, whatever that means.
The Transcript, which monitors earnings calls, reported that Constellation Energy projects the growth of AI usage will severely strain electrical generation and transmission capacity. This will slow and constrain the rate of AI adoption.


AI bellwether NVIDIA faces its own challenges. The company’s Q4 2023 10K report estimates that 19% of its sales come from a single customer.


In fact, NVIDIA’s sales are highly correlated with the capital expenditure rates of the Magnificent Six (Magnificent Seven ex-NVIDIA).


NVIDIA’s sales have gone parabolic, even by dot-com bubble standards. The accompanying chart shows NVIDIA’s sales compared to Cisco Systems, which was an internet infrastructure darling of the dot-com bubble. The key question then becomes, “What happens when either the rate of AI adoption is slower than expected because of the factors I’ve cited?”



The bull case

While I acknowledge the risks to the AI hype, here is the bull case for AI stocks.


If the concern is over earnings adequacy, current earnings levels appear reasonable, at least for AI bellwether NVIDIA. The previous chart compared the evolution of the sales of NVIDIA and Cisco Systems, but NVIDIA’s profit margins dwarf Cisco’s in magnitude.


In fact, today’s technology stocks are earning a far greater differential in profitability versus the S&P 500 than the dot-com era.


If the concern is over excessive valuation, the S&P 500 forward P/E is elevated, but it’s not at bubbly levels. If you buy into the AI hype that it’s a transformative technology, the bull has further room to run.


If the concern is over excessive bullishness among investors, I would suggest that today’s stock market is not frothy at all compared to the dot-com bubble days. Margin debt, which is an indirect indicator of individual investor speculation, hasn’t kept pace with the advance in stock prices.


The top of dot-com era was marked by a flood of low-quality IPO financings and eye popping price surges by new IPOs. The relative performance of today’s IPOs shows that the animal spirits of 2000 are not present.



For once in my miserable life…

In conclusion, well-known value investor Jeremy Grantham recently sounded warnings about an AI bubble. While I respect Grantham’s views, he suffers from the value investor problem of being too early and overly reliant on valuation for his views.


Here is an example. Just before the market bottom in 2009, when the S&P 500 was at 676, Grantham wrote that he believed fair value was 900 and he revised his forecast during the summer of 2009 that the index could rise as much as 1100. Those were correct calls. At the same time, he was forecasting a bubble in emerging market stocks and acknowledged to value investor’s curse of being too early. While EM stocks did soar, the real bubble was in FANG names.

For once in my miserable life, I would like to participate in a bubble if only for a little piece of it instead of getting out two years too soon. Riding a bubble up is a guilty pleasure totally denied to value managers who typically pay a high price to the God of Investment Discipline (Thor?) for being so painfully early.

Today, the market is starting to blow a bubble in AI-related plays, but I think it’s far too early to declare that a top is near. AI profitability is stronger than the stocks of the dot-com era. As well, investor psychology and the hype cycle haven’t become sufficiently frothy for investors to turn cautious.


If for once in your miserable life you would like to participate in a bubble, this is your chance. In the short run, large-cap technology stocks may be about to undergo a pullback or correction. Investment-oriented accounts should take this as an opportunity to buy the dip.


9 thoughts on “A reply to Grantham’s AI warning

  1. In your view (and I realize you don’t give investment advice) what names would you focus in on during this pullback? Or what ETF? I fit Grantham’s discipline to a tee, so I haven’t been in this AI rally. I was looking for a way to get some exposure and diversification through an ETF?

    1. Ken’s idea of momentum is a reasonable one. If you are looking for more direct exposure to AI, consider the semiconductors like SMH, or just QQQ for Big Tech exposure.

    2. Mike, I’ve had good results from QUAL for the past few years. Also check out DYNF, which I just learned about this week.

  2. I am skeptical.
    The most important question for me is how they will monetize this investment in AI. There is a similarity to interest rates…when people fear borrowing, lowering the rates is not effective. Actually you likely just get desperate at risk of default people borrowing, this is speculation on my part.
    Well, if people don’t spend more money, or even less because of the economy. How do they monetize this?
    True that AI could make the worksite more productive but likely at the cost of jobs. How much waste can AI prevent?
    The internet is great, but for many small and not so small businesses it has been a disaster as online businesses have killed them off. But in the aggregate are we richer?
    Maybe AI will somehow get us to reduce the cost of commodities, or help with robotics…will that be worth trillions year after year? Dunno.
    Money flows matter. If money goes pouring into one space, prices go up until it starts to rush out…baby with the bathwater happens in those times.
    A recent example of things is bitcoin halving. Less coin means more scarcity right? So prices will go higher, right? Well consider that BTC has a Ponzi element to it in that it feeds on itself. You can’t trade bitcoin without the miners. One buys bitcoin as a speculation store of wealth or sell higher or for criminal activity. It’s rare that you can get clothes or a pizza or car repairs with it. It is not systemic to the economy like Visa.
    So here we have a miner, whose resources have just dropped 50% in yield, where AI demand for electricity is unlikely to lower it’s cost, more likely electricity goes up in price, and AI is unlikely to make BTC more secure as the computing power increases. What happens to mines that are not profitable? Take a copper mine, it might keep cranking out the copper as long as the marginal cost of production is below prices but at some point it gets put on maintenance or shut down. This is not allowable for BTC. There is no inventory of BTC like copper stocks that traders can sell because BTC cannot trade without the miners. So what happens if we get a crash and large selling of BTC happens? Enough that the cost of electricity is more than the coins are worth?
    Will the Fed rush in to save BTC? Not impossible seeing how the world is, but my bet would be no. So, would you buy that kind of mine?
    AAPL could make some money out of AI if it comes up with some kind of iPod-like multi language translation device that works better than google translate, but how big is that market? It would be big but for most people not that useful since most people don’t go off to foreign countries.
    Didn’t people during the dutch tulip bubble know that tulips reproduce and people would be growing them like crazy?
    So I won’t chase BTC or NVDA…the easy money is done so to speak.

  3. AI adoption this time around will not have a big bubble. With big data and analysis tools so advanced today and getting better everyday, market will figure out winners and losers in real time, not wholesale movement we saw during dot com bubble. Friday saw Adobe dropped by 14%. Adobe is one prime target AI tools will attack. Similarly Canva, a pre-IPO company from AUS, will see its valuation dropped. One company WIX will also be affected although its share price still has not shown it.

    Even inside SMH if you look at performance of member companies, only those related to AI and data centers are moving. But AI tools already have a very big impact on those software companies in coding and testing. And then the entertainment industry is really in a bind. I feel that a lot of established companies will be affected greatly and you will see a lot of small outfits established to compete for the jobs. In music industry it will get much much worse.

    A lot of real high tech companies in various fields really benefited from AI tools in idea generation. So yeah, a lot of work to figure out where to put your investment money. The world is getting more and more concentrated. Again Pareto Principle.

Comments are closed.