The planet’s most important stock, and most valuable company, topped estimates in quarterly results released after the bell on Wednesday.
Nvidia rang up $35.1 billion in revenue during Q3, up 93% YoY and easily ahead of the $33.25 billion consensus.
I say “easily.” That suggests the top-line counted as a big beat, but at this point, a year and a half on from Jensen Huang’s “guide heard ’round the world,” it’s very difficult to discern the location of the bar to clear, shrouded as it is by all the clouds up there.
Huang sounded his euphoric self. “The age of AI is in full steam, propelling a global shift to Nvidia computing,” he declared.
Nvidia guided for $37.5 billion in sales for the current quarter, plus or minus 2%. That was “merely” in line.
In the color accompanying the release, Huang gushed about Blackwell. The anticipation’s “incredible,” he said. I don’t know if that’s a good enough superlative. Previously, Huang used “insane” to describe demand. Later, on the call, Colette Kress tried out “staggering.”
The breakdown — which is more or less irrelevant now unless you want to argue anyone’s going to buy or sell based on the gaming readout — showed data center sales were $30.8 billion against expectations for $29.14 billion. Adjusted gross margin was 75%, in line, and non-GAAP EPS of 81¢ was a beat. The current-quarter margin guide matched estimates at 73.5%.
I’d be lying to suggest I have special insight into the Blackwell question(s), and frankly I think that puts me in good company. Most of the Street was flying more or less blind on that headed into results, and notwithstanding what’s guaranteed to be a lot of colorful language from Huang, the in-line revenue guide perhaps suggests the trajectory of Nvidia’s revenue acceleration is becoming more predictable. That’s not necessarily a bad thing, but the days of multi-standard deviation beats may be over. For now.
Huang carried on about the revolution he’s leading. “AI is transforming every industry [and] company,” he said, before talking up robots (“physical AI”) and country-level initiatives as governments “awaken to the importance of developing their national AI and infrastructure.”
It’s possible that some investors will view the results as a small letdown, but I think these numbers are more than enough to keep the dream alive. While emphasizing that I’m not an authority on Nvidia by any stretch, I don’t view Wednesday’s relatively mundane release as any sort of disappointment.
Readers can correct me if I’m wrong, but given that Nvidia’s customers are surely waiting on the full-scale Blackwell rollout to buy, the solid top-line beat for Q3 and in-line outlook seem like a good outcome, particularly given what I imagine are pretty high odds that Huang ends up easily beating that $37.5 billion current-quarter guide.
Personally, I wouldn’t sell into this report. But with the shares up — checks notes — 200% in 2024, I’m not sure I’d be buying into it either.
The muted post release reaction is probably due to the MASSIVE options positioning going into the number. As you’ve tutored us, this sort of options positioning can act as a vise around either side of the strike price before the options expire.
Maybe those writing these options finally came up winners for once.
Longer term, the need for paying end users to run a LLM relevant to their business needs is increasing a “one and done” thing which does not require ongoing access to vast Nvidia GPU arrays. Much cheaper non-Nvidia edge computing where Nvidia lags AMD and even the hated INTC offer better suited chip sets.
But that’s an argument for another day down the road.
Nice placement of the Nvidia ad.
Nice post-earnings analysis Sir. Much Better than the babbling on CNBC etc. As long as TSMC can increase supply and Micron and Hynix can increase HBM memory supply then I think NVIDIA should grow consistently for next few yrs. and the stock should have less volatility. Their balance sheet is rock solid !!
I will be forever grateful to Jensen, and before him- Hock. I am searching for another company (profitable now) that looks undervalued today, relative to potential over the next 5 years. I am still finding it difficult to look too far outside the technology sector.
Over my academic career I took a dozen or so econ courses and earned a couple degrees from that immersion, I don’t really consider myself to be an economist because as a group they understand little about how businesses actually work. However, I do remember some of the ancient orthodoxy, namely the field called Industrial Organization. That’s the study of the substructures firms form as they evolve. The four main choices are pure competition (no one there makes any real money), monopolistic competition (few examples), oligopoly (the smallish focused home of most of the country’s serious firms, two or three are bosses), and monopoly (mostly illegal and giants making much money). The way much of this this works is that small firms, or groups within a larger firm, start with an innovation. If it’s any good the firm/group may grow, fast even. If the innovation is disruptive it will attract attention. At this point, long-term success will depend on the ability of the innovator to protect their creation. If the heart of the new creation is an idea or is not particularly novel, it can’t be patented. Even if it can be protected, it won’t be for long. There will be competitors! If lots of capital is required, competition will be limited and an oligopoly will probably form, or the source of the idea may be acquired by large capital suppliers. AI is not novel and it is only an idea so it can’t be patented. Anyone wants to can get into this business. Most chips and software are hard to patent, though most can be copyrighted, so substitutes are possible. It only took three of four years for AMD to duplicate the operations on Intel’s first microprocessors. So, what? Nvidia will grow, but larger competitors will appear, force down margins and offer average industry profits. Successful and unsuccessful AI applications will be created and eventually the firms creating the real deal applications will grow faster than NVDA and its competitors. NVDA makes tools to be used in various (AI) applications. But other firms will make different tools. AI itself will be bigger than all of that. It will be the applications overlord and define the domain. Investors will be the ones who will have to guess who will be in charge of what and how long any of that will last. Me, I’ll be gone just like Baldwin Locomotive that died trying to sell annuities.