Nvidia Just Cleared Its May Record, and the Street Still Wants More

$237.83. That's the intraday high Nvidia hit on October 2, and it's the first time the stock has traded above its May peak.

For a company that spent most of the summer going sideways while skeptics sharpened their knives, that's not a small thing. The immediate catalyst wasn't a chip launch. It wasn't an earnings beat. It was a note.

Morgan Stanley's Joseph Moore restored Nvidia as his firm's top semiconductor pick, and he did it after meetings with CEO Jensen Huang and CFO Colette Kress. No new product, no guidance raise, just a conversation and a ratings call. And yet the Street's most bullish price targets still imply upside of up to 99% from here, which raises a fair question. What exactly are those analysts seeing that the market hasn't already priced in?

Why One Note Mattered

Sell-side ratings don't usually move a company this big. Nvidia is the largest listed company on Earth by market cap, covered by dozens of analysts, and generally immune to a single upgrade. But Moore isn't a perma-bull, and the fact that he restored the top-pick slot implies he'd stepped away from it at some point. Analysts don't reverse themselves on a whim.

The meetings matter more than the rating. Moore went in looking for cracks in the demand story, and apparently didn't find them. That's the whole thesis in one sentence. The orders are still there.

So the stock broke out. Granted, the volume behind the move was decent rather than spectacular, and one record day doesn't erase three months of chop. But it does tell you where the consensus sits going into the next earnings cycle. Nobody wants to be the analyst who blinked first.

The Problem With 99% Upside

Here's where I get uneasy. That upside number isn't derived from Nvidia's products. It's derived from the spending plans of a handful of hyperscalers, and those plans are enormous, concentrated, and increasingly interconnected with the companies supplying them.

Color me skeptical, but a bull case that depends on the same small group of buyers funding both the demand and the capital behind the supply chain is a fragile thing. It works beautifully until it doesn't, and the unwind wouldn't be gradual. History suggests otherwise on that front, admittedly, but the concentration here's unusual.

To be fair, the proponents have a point. Data center buildouts take years, not quarters. Power constraints are real, and they keep demand rationed rather than collapsing. The competitive picture is messier than it was in 2023, with custom silicon from the big cloud players taking a slice, but Nvidia's software stack still does the heavy lifting for most serious training workloads.

So the narrative holds. What I'd push back on is the idea that a record high plus a bullish note equals a clear runway. It doesn't. It equals a stock priced for a lot of things to go right at once.

What to Watch

Forget the price targets for a minute. The number that actually matters is hyperscaler capital expenditure guidance, and you'll get fresh reads on that over the next few weeks. Nvidia's own November print is the other marker. If the company guides anywhere near the more aggressive Street models, the 99% crowd looks smart. If it guides conservatively, expect the same analysts to explain why their target was always a multi-year view.

The question worth asking: is this a company still early in its cycle, or a company whose stock has already absorbed three years of good news into one price? I'm not entirely convinced it's the former. But I've been wrong about Nvidia before, and so has almost everyone who bet against it.

Boring answer, I know. Records tend to make people forget that.