One Bank Asks: Is The Tech Ecosystem Now Its Own Asset Class?
People are beginning to ask some -- how should I put this? -- interesting questions about equities and specifically about tech dominance.
Last week, ahead of Nvidia's earnings release, there was talk of the "S&P One," an allusion to the (not completely misplaced) notion that the US equity market is really just one stock at this point.
Laugh as you will, but have you seen the return attribution for the index? If not, and lopsided dynamics bother you, I suggest you avert your eyes.
In his l
NVDA is a hardware company, and hardware companies almost always fall to the margin-crushing weight of competition. For an excellent example of what happens, just look at the history of Nvidia’s own industry. Chip making is a competitive sector with a lot of players. They (Nvidia) are enjoying the absolutely remarkable benefit of arriving with exactly the right product at exactly the right time and shipping for size, but it just takes one technological advance by one of their competitors, and it rapidly dries up. But I’m sure gamers and crypto miners will buy their products. At a discount.
The only way a hardware company can stay on top is if they have a brand monopoly. There’s an excellent example of just this: perhaps the greatest brand monopoly of all time, Apple. People will happily pay several hundred dollars extra for Apple products. No one else can make an Apple product. Sure, you can make a phone that’s every bit as good–better even–and sell it for less, but it’s still not an Apple.
Moreover, hardware companies have huge manufacturing overhead. You have these wildly expensive machines, an army of highly skilled workers to operate them, and wildly complicated supply chains. I mean, TSMC is trying to build a chip fab out in Arizona, but the problem they ran into wasn’t finding a way to supply ample water to a thirsty manufacturing process in the middle of a desert, it’s that they can’t find enough skilled workers. Meanwhile, software companies’ cost to manufacture another unit of product is a trivial amount of electricity and bandwidth. The most expensive part of Microsoft manufacturing another copy of Excel is processing the credit card transaction.
This is why nobody talks about IBM anymore. It’s easy to forget that at one point, IBM was the largest (by market cap) publicly traded corporation on planet earth. But it turns out anyone can manufacture a PC. The moat was the software, and they outsourced that to some dude named Gates who didn’t even own a tie.
I was listening to a podcast once, and they were cheering one of their members finally buying an Apple so that everyone in the group chat had the same color text bubble. No one talks about chips like that unless they’re in an industry where processing power is king. Even then, while a professional gamer, animator, or AI engineer can tell you all about the chips in their hardware, all they care about is performance: GHz, threads, cores, power consumption… They’ll drop their current supplier in a hot second if something better comes along.
Incidentally, this is the same reason Tesla is doomed to revert to the mediocrity of the mean. They did have some brand loyalty for a hot minute, but Elon squandered most of that in order to curry favor with a crowd largely unable to afford a Tesla, to say nothing of a crowd that often actively rails against EVs for some reason. But I digress.
There’s no reason Nvidia’s run of AI chip dominance can’t continue for another 18-32 months. They’re going to make great heaping piles of cash. But as soon as they get some competition, the party’s going to end and the hangover will begin. Don’t worry though, there’s no doubt in my mind that they’ll at least fare better than IBM.
My family’s smallish business took off in the late 1930s when my grandfather developed the “stove lamp.” Kitchen stoves had no lighting and there were no hoods in those days so it was difficult for housewives to see what they were cooking. Then came us, with a bright light that could be custom configured to attach to any stove on the market. There were over a 100 stove makers in those days and we sold our lights to all of them. And we had no competitors. What a moat … for a couple years. The trouble is that it was such a good idea that the major stove companies soon redesigned their products to build in needed lighting and eliminate us.
If AI becomes as important as it seems, the same thing will happen to Nvidia. Google has been designing producing their own custom data center chips for some time. Amazon has developed some amazing robots for their logistical needs. I am not an industry expert but I’m pretty sure Microsoft does plenty of its own work, along with Apple. And of course, Intel is still around. Nividia’s people aren’t chained to the wall and can move. Products are the hardest moat to protect in business. Virtually anything can be copied. AMD proved that many years ago. Innovative processes are easier to keep secret but most of the hardware processing innovations are developed by foundry equipment makers and suppliers. Nvidia has been clever and gotten ahead of many players, for now. They will be had if there is real money at stake, and there is certainly that. Long ago when IBM developed and introduced the first PC, the division director was asked why IBM didn’t come out with personal computers before Apple and others. Why wasn’t IBM the pioneer? The IBM guy replied, “Do you know what a pioneer is?” “It’s the guy with arrows in his back.” One of my all time favorite lines.