Amid the Anxious Nvidia Earnings Watch: Losing the Plot, Misreading Market Dynamics

I sometimes wonder whether COVID and its aftermath have unmoored us from what we used to collectively apprehend as our common reality.

Take, for example, the intense media coverage of Nvidia’s quarterly results, which are not due until close of market trade today. Journalists and market watchers are behaving as though we’re about to witness a blue moon, a suicidal comet, or the reunion of Oasis. The anticipation is, if you will allow me to say so, a little over the top, a touch frenzied. I am tempted to reach for the phrase derangement syndrome, but maybe that’s too strong. Then again, maybe not. These are strange times. 

MarketWatch (of the relatively staid Dow Jones business-media empire) is, believe it or not, running “live updates” on Nvidia’s impending earnings, and it has been doing so since about 6 am ET this morning. Usually one sees this sort breathless, minute-by-minute coverage only on grand occasions such as the World Cup, a heavyweight-boxing championship, or Apple product launches. MarketWatch is not the only exuberant offender, though it might be the most fervently infatuated. Everywhere you look online, or in the wider world of old-media screens and data, you see and hear discussion of Nvidia and what its latest quarterly results might portend; and not just for Nvidia itself but for the entire universe of technology companies and stocks. Some addled journos would even have you believe that the near-term fate of the economy hinges on what Nvidia releases to the world in less than two hours (I published this post before 3 pm ET on Wednesday, August 28, 2024).

These must be signs of an acute mania, no? Compelling circumstantial evidence that we have left the stolid station of rational thought and are now on the fastest of bullet trains, the g-force knocking us back in our seats, hurtling instantaneously toward the exotic, albeit illusory, destination of magical thinking?

Wheels in Motion 

Here’s why I think the minute-by-mine live update and the hyperventilated anticipation are signs that the technology industry, inclusive of its media enablers, has taken a wrong turn. If we’re not careful, we’ll soon be forced to endure the tech industry’s answer to Entertainment Tonight. Maybe we’re already there.  

Despite what some might see as my dour long-term view of Nvidia’s prospects, expounded at length in this forum on more occasions than I care to admit (including one salvo very recently), I am also of the view that Nvidia, in the here and now, will likely beat its consensus quarterly numbers for revenue and earnings. Many of its largest customers, namely the hyperscalers, are still ordering and procuring Nvidia’s GPUs and AI accelerators. 

Oh, the wheels are in motion and the change is afoot, but major shifts in technology infrastructure do not happen overnight. Designing, developing, and building infrastructure takes time, as does the transition from what it is deployed today to what will be deployed in future. It’s a major endeavor, with most of the progress occurring behind the scenes in imperceptible increments. We don’t see it happening; we only see the results, which take time and arrive at a later date. 

That said, whenever Nvidia’s ascent begins to plateau, when its revenue judders and then sputters, its profits becoming less certain, we should resist the temptation to interpret the company’s waxing or waning as a definitive proxy for the industry as a whole. Nvidia inevitably will struggle, at least partly for the simple reason that it enjoyed too much success, causing its largest customers, the cloud giants, to reach a point where they could neither countenance nor afford (and I use that last word with great care) to buy and depend upon Nvidia’s products and its corporate goodwill. 

When that day comes, when the hyperscalers finally unyoke themselves from an increasingly costly dependence on Nvidia, the travails of the latter won’t be attributable to the struggles of the cloud giants, who will continue to develop new services, including extensive genAI offerings, and will continue to grow their globe-girding empires. In fact, the cloud giants won’t struggle at all, even as Nvidia hits a wall on Wall Street. 

The success of Nvidia is not ineluctably and reciprocally linked to the success of the hyperscalers, or even of enterprises who use the cloud. We have misinterpreted events, assuming incorrectly that the successes of Nvidia and the hyperscalers are mutually reinforcing and mutually dependent. That’s not true. If you’ve paid close attention to the business dynamics and pressures the hyperscalers have described for the past year or so, you will know that a split is coming, likely amicable and gradual, at least initially. 

When it came to playing a certain game, which involved the provision of GPUs for genAI, Nvidia had the rarest of otherworldly runs; it was practically the only game in town. Those circumstances, however, are not sustainable. Eventually, in any vital market, competitors emerge, presumably impermeable moats are systematically eroded, and exceptionally deep-pocketed customers (yes, the hyperscalers) choose to reassess the buy/build/partner analysis and decide it’s time to build their own technology rather than rely, at great monetary and strategic cost, on a third-party. 

This isn’t a new chapter in the history of information technology. What I’m describing is not, in any way, revolutionary. It’s how markets work. Nothing lasts forever, and sometimes, in the technology business, one is lucky if something lasts more than a few years. 

False Reciprocity 

Let me restate my key point more strongly: Nvidia’s impending quarterly results – whether exceptionally good or slightly disappointing, now and in the quarters to follow – should be perceived as indicators only of Nvidia’s business prospects, not those of the wider technology universe or the economy as a whole. Nvidia will ultimately meet the fate of every other vendor of “picks and shovels” in the cloud era. There’s nothing about Nvidia’s circumstances or fundamental profile to suggest that it controls or dominates the practices or buying behavior of its current hyperscale customers. 

Just as Cisco’s slow but steady decline as a purveyor of datacenter switching in the early years of the cloud era did not portend the demise of networking in cloud datacenters – there are mountains of non-Cisco switches in the datacenters of the hyperscalers – any decline in Nvidia’s GPU standing in the datacenters of the cloud giants will not signify, in and of itself, the death or terminal illness of genAI within hyperscale datacenters. Instead, what such a decline in a single vendor’s fortunes will signify is that the hyperscalers, individually and collectively (but each in its own way, for purposes of differentiation) chose to replace Nvidia’s chips and systems with their own or with those of a Nvidia competitor. That’s all it will mean. There’s no bigger-picture story to worry about.  

This is also why Wall Street analysts should exercise caution when assuming a direct relationship between the capital expenditures of hyperscalers and the revenue growth and profitability of Nvidia. 

Why? Nvidia’s share of those capital expenditures might change, for the reasons I’ve discussed here many times before. Past performance, as the timeworn axiom decrees, is no guarantee of future results, and just because something was true in the past, one should not assume that will remain true into perpetuity. Sometimes the very fact that something was able to materialize as a market reality, that it achieved success on a massive scale, served as the cause of its downfall, akin to a Shakespearean tragedy but with less pathos and, one would hope, less bloodshed. 

I expect – and, in genuine humility, I am but one member of a large crowd holding a similar opinion – Nvidia to post strong quarterly results later today, including modest or immodest beats on revenue and profitability. I also expect Nvidia to offer relatively encouraging guidance for its current quarter.

At some point, though, the narrative will shift as the hyperscalers reassert their prerogative – of their capital expenditures, of their genAI product differentiation, of their product-release schedules and differentiation – by adopting homegrown GPUs and AI accelerators. As such a shift occurs, Nvidia’s results are likely to be adversely affected, but I fail to see how such a scenario occasions an industry-wide malaise.

We might find ourselves suffering from the economic effects of a genAI hangover, if the technology’s quantifiable benefits fall short of its bombastic promise, but that will not be because of an existential linkage between the fates of Nvidia and the hyperscalers.