Fall is Upon Us, but Tech Markets Feel the Heat

These are crazy, nervous times. Derangement and irrationality, like two mad dogs, have strained against and broken free from their leads. They’ve run from the yard, chasing panicked passersby down the street, snapping wildly at their heels. What can you do but watch and hope nobody gets hurt?

The latest “meme” stock, Beyond Meat, has skyrocketed in value, completely uncoupled from gravity and underlying fundamentals (as if those still matter). Investing in meme stocks is like playing a game of musical chairs, but if you’re not listening closely, you’ll end up deprived of money and a place to sit. It’s not a game for the circumspect, nor is it the sort of investment made by anyone who doesn’t intend to spend their waking hours fixated on a stock-trading screen.

Meanwhile, gold goes up sharply and steadily, then takes a sudden drop, only to regain its stride a couple of days later. The vertical ascent of many AI-related tech stocks continues to defy reason. All the while, though, suspense builds. Nearly everybody expects a plot twist. The narrative demands a climax before a denouement, which will provide a happy ending for some, a sad ending for others, and indifference for everybody else. We can all agree that we’re following the story of AI, everybody has an opinion about how it might unfold, but nobody really knows what lurks around the next corner.

You can feel the tension out there. Angst is pervasive, seeping through the economic, political, and societal spheres. Indeed, angst might be a word whose origins are Danish, Norwegian, Dutch, and German, but the universal foreboding it describes has sped like a rocket-powered hovercraft across the pond and now squats sullenly in our midst.

Even among the most ebullient AI optimists, whose sunny dispositions occasionally obscure a shadowy self-interest, concern mounts amid a swirl of mutually reinforcing circular financing pacts. Meanwhile, the looming specter of a private-credit crunch has already squashed a few stray “cockroaches,” in the indelicate phrasing of JPMorgan Chase’s head honcho, Jamie Dimon.

Everybody in the AI vanguard, from Nvidia to OpenAI, gives the impression that the extravagance and bacchanalia will go on indefinitely. Perhaps it still has a ways to go, depending on your sense of time or your degree of equanimity. Just how concerned are you? If you’ve already taken ample profits, you’re probably at peace with the world.

How Different Is It, Really?

Some people suggest “it’s different this time,” but forgive my skepticism. I’ve lived through several market crises, and on every single occasion the charlatans and the unwary said the same thing: It’s different this time. It’s always different, until it isn’t.

Optimism is fine, I have no real objection to it, but one needs to guard against blind spots. If you don’t check for blind spots, you get t-boned by a hideous Cybertruck, suddenly unself-driving and veering across from the passing lane like a misfiring rocket a Cape Canaveral.

Yes, there’s an opportunity cost imposed on the circumspect, but that’s really the only cost they pay. Yes, the cautious miss out on capitalizing on further upside in a bloated, wobbling market. It’s true, too, that it’s only in retrospect that anybody can truly call the top of a market. If we had the prescience to know when we’d reach the top of the market, the descent would be graceful, not jarring and precipitous. The fact is, the drops, when they come, are rarely dignified. For some, the losses are staggering, the undergarments often soiled.

This time, as in so many others, the losers will be those least equipped to take the hit. The largest technology companies, including all the hyperscale cloud providers, will survive, though at least one might look back on the approaching market dislocation as a fateful turning point in a permanent corporate decline. I don’t know which hyperscaler that might be — a couple of suspects come to mind — but there’s always one giant that suffers permanent impairment after these market temblors. Meanwhile, many currently up-and-coming AI startups will implode in the downdraft, while one or two will get promoted to the big leagues.

As it stands, the AI mania is driven by sentiment and its close relation, momentum. We prize rationality, but it’s rarely in the market’s driver’s seat. At this point, I suspect it’s even given up on backseat driving.

Billions of Inferences Served

As we’ve discussed, datacenter buildouts proliferate unabated, growing in size and number by the week. The announcements are literally thick on the ground. These datacenters are being built to serve anticipated future demand for AI services. I have reason to believe that some of that demand will materialize, but datacenters are multiplying faster than McDonald’s franchises in the 1970s.

Is it rational to build datacenters so aggressively and abundantly, at a rate that cannot help but end in excess supply? Or is this buildout a function of those toxic twins we have come to know as fear and greed? (The first two interrogative sentences in this paragraph are known in literary salons as rhetorical questions. Technically, given that I asked one question followed by another to reinforce my point, I suppose I’ve invoked a particular type of sequential rhetorical device called a pysma, though mine is admittedly a crude vintage. Plonk-fermenting class is concluded for the day.)

Actions beget reactions, so perhaps it comes as no surprise that a countervailing force has risen to meet the datacenter boom. A digital-age Nimbyism, discussed here previously, manifests in communities that would rather not accommodate unprepossessing, windowless, noisy, energy-guzzling, water-gulping behemoths as their new neighbors. If we’re honest, we can understand those objections. We nod in sympathy.

Nonetheless, regardless of oppositional resolve or vehemence, public resistance to mushrooming datacenters will not be why this AI market mania goes pear-shaped. The community activists will merely complicate corporate decisions and stagger timelines relating to where datacenters go and how fast they get built. (There’s an entirely different post I could write about how the technology industry, once viewed by the masses as a largely benign presence, is now increasingly perceived as a malevolent leviathan. The industry has only itself to blame for this change in public perception.)

A bigger threat to the construction of AI datacenters and their associated infrastructure is represented by the trigger-finger fickleness and inconstancy of the market, composed as it is of an array of investors great and small. A significant portion of the frenzied datacenter buildout is financed indirectly by the rising valuations of publicly traded technology companies, namely hyperscalers. Yes, the revenue growth and rude profitably of Google, Microsoft, Amazon, Meta, Apple, Oracle, Nvidia, and others helps feed the fire. Yet, it’s the promise of future riches, reflected in the stock prices of the aforementioned companies, that really keeps the party buzzing.

Diversity and Moderation: Not Exciting, but it Works for Me

As mentioned at the beginning of this post, investors are anxious. They study P/E multiples and break out in a flop sweat. Various pundits tell them a reckoning is at hand, a correction imminent, a bubble about to burst. Nobody knows exactly what will happen, or how bad it will be — and, of course, nobody knows precisely when it will all go down — but the mortal dread is undeniable.

Nobody wants to get caught on the wrong side of the market’s fault lines. People can do the math, they can readily ascertain that the best time to get into Nvidia, for example, was a few years ago, not now. There might still be upside to investing in Nvidia, even at its current valuation, but there’s a lot of risk in the gambit. Considerable optimism is baked into the valuation, and not just for Nvidia.

I’m willing to leave that trade alone. The fear of missing out is restrained by the fear of losing my shirt. I wish I’d bought Nvidia in 2017, or even 2020, but the technology industry has not delivered reliable time travel and I can’t go back. There’s no point in even looking back. What’s done is done, and we have only the present unfolding into the notional future.

As I look ahead, trying to focus my foresight on that oblique future, I see danger in the form of a sharp correction, perhaps even (though less likely, as I think there’s some substantively value in AI) a bubble that bursts and leaves a years-long trail of detritus. I’m still traumatized, you see, by the psychic scars left behind by the bursting of the dotcom bubble. Everybody mocks Pets.com, but perhaps we forget that there were plenty of real companies, with balance sheets and assets, that were severely damaged, some permanently, by what occurred in the early years of this millennium.

I’ve learned hard lessons that only experience can teach. You say, well, it’s different this time. Maybe it is, but I’m not willing to put my money on it.

The best advice I can give, informed by an eventful lifetime of sentient experience, is to practice diversification and moderation in all things, including your investment portfolio. Don’t get greedy, don’t be steamrolled by fear, but be careful out there.