An analysis of Nvidia's earnings report

AI-sector bellwether Nvidia (NVDA) reported second-quarter earnings and boosted guidance after the close yesterday (here's the press release and investor presentation).

There's no sign of a slowdown in demand for its chips, which are at the heart of the AI-infrastructure boom...

Revenue soared 18% from the previous quarter and 106% year over year ("YOY"). But it came in only 4% above expectations. And revenue guidance for the third quarter was only 3% above expectations.

Adjusted earnings per share of $2.22 jumped 19% from the previous quarter and 120% YOY. But again, this was only slightly above estimates of $2.09.

In light of these great – but not blowout – numbers, the stock initially fell a few percentage points. But the stock popped as much as 8% this morning because on the conference call, the company massively boosted guidance...

It projects that revenue will grow a staggering 70% in the next fiscal year ending in January 2028, far above expectations of 45%. And it suggested that growth would be more than 100%, if not for ongoing supply constraints.

My friend Scott Tashman of Outset Global Trading addressed the call in an e-mail this morning:

Nvidia earnings calls are usually very bullish, but last night's one was particularly effusive. Even more important than the 70% number were various remarks from CEO Jensen Huang and [Chief Financial Officer] Colette Kress about certain aspects of the AI industry: AI is now both extremely useful and increasingly profitable, hyperscalers and neoclouds are earning substantial sums on their AI compute infrastructure, artificial general intelligence has been achieved, agents recently surpassed humans in token consumption, etc.

However, he noted that the earnings release "wasn't all roses":

Revenue in AI clouds, industrial, and enterprise segment fell short, days sales outstanding spiked quarter over quarter (which feeds into worries about the company manufacturing demand by offering generous purchasing terms to customers), and the gross margin forecast fell short (due to extreme memory cost pressures).

If Nvidia is witnessing margin headwinds because of memory, it's safe to say everyone else in the industry is getting hit even more (Nvidia recently raised its own prices and has the best memory procurement deals of anyone – if they aren't immune, no one is).

As for the implications in the broader AI sector, Scott wrote:

The 70%-plus sales guidance for fiscal 2028 implies industry capital expenditures ("capex") could accelerate far beyond expectations, and while Jensen was very upbeat on token economics, the industry's debt issuance/free cash flow problem will probably get worse over the coming quarters (which will place substantial upward pressure on yields).

The market for months has toggled between AI pick-and-shovels and hyperscalers/software, based on the view that strength in the former would hurt cash flow for hyperscalers and disrupt software. However, this relationship is starting to break down. To the extent token economics are improving by as much as Nvidia suggests, the hyperscalers and neoclouds are well positioned to benefit.

This Wall Street Journal article also captures some factors investors are worried about:

Nvidia is writing large checks to its suppliers even as it amps up financial backstops for less-creditworthy customers. The strategy might help the company skirt supply constraints and realize lofty revenue growth forecasts, but it isn't risk-free.

Nvidia said Wednesday that its commitments to suppliers to secure components for its AI chips and systems reached $279 billion in its latest quarter, more than doubling from $119 billion the previous quarter.

Nvidia's supply-chain commitments have spiked by more than 6 times YOY, as this chart from the article shows:

As the article continues, this could end poorly for Nvidia if (or when) the AI bubble bursts:

While commitments can grease the wheels of a supply chain during a boom, they also serve as insurance for suppliers against an unexpected downturn in end demand. In such a scenario they can transform from being a strategic advantage for the company in the middle – in this case, Nvidia – into a painful liability.

That is precisely the situation that befell Cisco Systems in 2001. When the dot-com bubble burst, demand for Cisco's networking gear plummeted, but it was still on the hook for commitments to its own suppliers. It was able to negotiate away some of them, but took a $2.2 billion inventory charge in a single quarter.

More broadly, the article highlights Nvidia's growing list of financial risks:

Among them, Nvidia is providing a $105 billion backstop on an OpenAI data center lease and up to $125 billion of residual-value support as part of a $500 billion financing deal with Wall Street asset managers. It is also guaranteeing $36 billion in sales for cloud-computing companies it sells chips to, in exchange for revenue-sharing deals.

On top of these, Nvidia disclosed Wednesday that it has entered into data center lease agreements worth $20 billion that it expects to transfer to third parties. And it is providing "selective credit enhancement" for nearly two gigawatts of computing power to a "frontier AI lab" that isn't OpenAI, Kress said...

Frontier labs, Kress said, "are growing faster than what their balance sheets and credit profiles can support."

It isn't too hard to imagine a scenario where a bust squeezes Nvidia from multiple sides. Suppliers might pressure the company to pay up at the same time as financial commitments to suddenly less-eager customers come due. Declines in the value of Nvidia's equity investments, many of which are in its own customers, would only deepen the profit impact.

The article concludes:

Nvidia's financing and supply-chain moves might help it grow faster now, but magnify its risk longer term. The stock, already a concentrated bet on AI, is only getting more so.

My friend Doug Kass of Seabreeze Partners smells a "rig job" between Nvidia and AI firm Anthropic. In a missive this morning (for subscribers), he wrote:

Interesting that Nvidia apparently has not given revenue guidance a year out in ages, but now all [of] the sudden they give an annual +70% revenue growth number right in front of a theoretical Anthropic IPO filing?...

It's especially interesting in light of the fact that based on results (inclusive of the [days sales outstanding] spike and declining gross margins), the stock was trading off before they gave this year-out forecast, which really they have no idea, because nobody ever does. Their own history of blowing up tells you that.

I actually think the whole thing is something of a rig job between Nvidia and Anthropic, because apparently Anthropic is seemingly timing their IPO filing on the back of the Nvidia quarter as Nvidia needs Anthropic to be a public company to raise the money and help keep the shell game going.

Even Nvidia cannot finance the whole thing on their own and buy all of their own revenue, although they sure seem to be trying.

Doug poses an important question:

How does industry capex grow +70%? Balance sheets are shot, cash flows are negative, there seems to be no capacity for memory, no more power capacity, and states are now turning down data center requests. And then there are all the other industry issues that remain, including the move to open source and lack of underlying economics.

With all of this in mind, let's take a look at Nvidia's valuation...

Coming into earnings, analysts expected the company to earn around $9.05 per share this year and $13.13 next year.

So with the stock around $225 this morning, it's trading at about 25 times current-year earnings and 17 times next year's.

These are very low multiples for a company this dominant, profitable, and fast-growing. And if Nvidia hits its projected revenue numbers, earnings will be much higher than analyst expectations – and the stock could double in the next year or two. But that's a big "if."

The biggest risk to Nvidia is that ChatGPT owner OpenAI blows up, which would roil the entire sector. I think this is likely, for reasons I've outlined in many previous e-mails (archive here), most recently on Monday.

My concerns have only deepened after reading this in-depth cover story in the latest issue of Time: Inside OpenAI's Reboot. Excerpt:

It's been a difficult stretch for the company that ushered in the AI boom... Over the course of the past year, OpenAI lost the lead in the AI race to archrival Anthropic, which spotted the business opportunity in AI coding, built Claude Code into a market-defining product, and surpassed OpenAI in reported annualized revenue and private-market value for the first time. Anthropic, founded by OpenAI defectors, is now expected to be the first of the two companies to go public, two people familiar with its plans say, with the IPO as early as September...

As Anthropic surged, OpenAI suffered a series of setbacks, including a spate of leadership departures... Outside the company's revolving doors, challenges mounted. Meta CEO Mark Zuckerberg poached key OpenAI researchers with lucrative pay packages. Google's Gemini products now reach more than 1 billion people per month. Apple sued OpenAI, alleging theft of trade secrets. (OpenAI has denied the charges.) OpenAI battled its co-founder Elon Musk in a lawsuit accusing the company and Altman of betraying its nonprofit mission.

So what should you do if you own Nvidia's stock? I'll repeat what I wrote after its earnings report a year and a half ago:

[Nvidia] is the kind of stock I like to pound the table on when it's down at least 50% (if not 75%).

The stock isn't at those levels today. But in that same e-mail, I also noted that if I owned Nvidia, I wouldn't sell. (On September 5, 2024, I detailed how to handle this sort of "high-class problem" with a stock like Nvidia.)

As I've said many times before, you must let your winners run! But you might want to use a stop loss to protect your gains.

Best regards,

Whitney

P.S. I welcome your feedback – send me an e-mail by clicking here.