Nvidia $100 Billion Historic Quarter Confirms the AI Boom

Nvidia (NVDA) has done it again, delivering one of the most remarkable quarterly earnings reports I can recall.

Second-quarter revenue surged 106% year-over-year to $96.2 billion, with gross margins of roughly 75% and net income of $59.7 billion. Nvidia also returned $25 billion to shareholders during the quarter, while growth expectations for the coming year moved sharply higher.

The numbers themselves are extraordinary. But what makes this quarter historic is not simply the growth rate. It is the combination of growth and scale.

There have been other periods when enormous companies posted comparable growth, but nearly all came with important caveats. Oil majors roughly doubled nominal revenue during the 1970s, but much of that was driven by commodity prices and ultimately reversed. Pfizer nearly doubled from a $42 billion revenue base during the pandemic, but that was largely a one-time vaccine windfall. Amazon added $106 billion in revenue in 2020, but did so at a fraction of Nvidia’s profitability.

Nvidia is now approaching a $400 billion annual revenue run rate, equivalent to roughly 1.3% of US GDP. There is historical precedent for an individual company reaching that kind of economic scale, General Motors approached 3% of GDP in the 1950s, but it took GM 50 years to get there.

And that gets to the larger question surrounding the company and the AI boom.

Nvidia Is the Capital-Goods Supplier of the AI Boom

The best historical comparison for Nvidia may not be another fast-growing technology company. It may instead be the dominant capital-goods suppliers behind previous investment booms: locomotive and rail-equipment manufacturers in the nineteenth century, telecom equipment vendors during the late 1990s or the suppliers behind other massive infrastructure buildouts.

That distinction matters.

Nvidia's revenue ultimately depends on the capital-spending decisions of a relatively concentrated group of enormous customers. When spending is accelerating, the economics can be spectacular. But if those customers suddenly pull back, growth can slow very quickly.

Jensen Huang clearly understands that this is the central question facing Nvidia. That is why so much of the earnings call focused not simply on demand for GPUs, but on the economics customers are generating from them.

“Its tokens are productive and profitable. Now, compute is revenue.”

That claim is enormously important because it gets directly at the biggest concern surrounding the AI boom: Are companies actually making money on all of this infrastructure?

During the telecom bubble, enormous amounts of fiber were laid in anticipation of future demand, only for much of it to sit unused for years. If AI infrastructure were following the same path, Nvidia's extraordinary growth would look considerably more vulnerable.

So far, however, the evidence increasingly suggests otherwise.

Why GPUs Don’t Look Like a Dark Fiber Redux

Dylan Patel of SemiAnalysis estimates that compute can rent for roughly $13 million per megawatt, while leading AI labs can generate as much as $50 million per megawatt in revenue from that infrastructure.

Just as importantly, Patel notes that much of today's compute capacity is contracted before it is even built, which is close to the opposite of a speculative overbuild.

Only a year ago, many AI companies were losing money on every token they served. Today, inference economics have improved substantially, utilization remains extremely high and the largest AI companies continue to demand more compute than the industry can readily provide.

That does not mean the AI boom is without risk. If anything, the constraint may increasingly shift from demand to financing. Patel estimates the broader AI infrastructure buildout could exceed $11 trillion through 2029, with trillions potentially requiring outside credit. At some point, the sheer scale of that borrowing could push capital costs higher and slow the pace of investment.

But that is a very different concern from the idea that AI infrastructure is being built without customers or economic value behind it.

For now, the evidence continues to suggest that the GPUs Nvidia is selling are being heavily utilized, generating revenue for their owners and increasingly producing attractive economics. That makes today's AI buildout look considerably different from the dark fiber of the dot-com era, and gives Nvidia's extraordinary growth much stronger fundamental backing.

NVDA Stock Still Looks Attractive

Perhaps the most remarkable part of the Nvidia story is that, despite this growth, the valuation is not particularly extreme.

NVDA currently trades around 23.5x forward earnings and carries a Zacks Rank #2 (Buy).

Following a quarter like this, I would also expect analysts to continue raising earnings estimates, potentially providing another tailwind for the Zacks Rank.

The technical setup is equally encouraging.

NVDA shares have spent much of the last year moving through two major consolidations. As shown in the chart, the stock is once again pressing against resistance near its previous highs and appears to be attempting another breakout.

With Nvidia now providing further confirmation of both the extraordinary demand for its GPUs and the profitability being generated by AI compute, a successful breakout from this consolidation could mark the beginning of another meaningful advance.

Image Source: TradingView

Should Investors Buy Nvidia Stock?

There are still legitimate risks, as there always are.

Nvidia depends heavily on an extraordinary capital-spending cycle among a relatively small number of customers. Financing requirements are becoming enormous, memory costs could pressure margins and eventually the economics of the AI buildout will have to justify trillions of dollars in investment. But the evidence available today continues to move in Nvidia's favor.

Demand remains exceptional, utilization is high and AI infrastructure increasingly appears capable of producing attractive economics for its owners. Nvidia's growth remains almost without historical precedent, and yet the stock trades at a valuation that looks surprisingly reasonable relative to that growth.

The AI investment cycle will not continue at this pace forever. No capital-spending boom does. But Nvidia's latest quarter offers little evidence that the cycle is approaching its end.

For now, it remains difficult to make a compelling case against owning NVDA.

7 Best Stocks for the Next 30 Days

Just released: Experts distill 7 elite stocks from the current list of 220 Zacks Rank #1 Strong Buys. They deem these tickers "Most Likely for Early Price Pops."

Since 1988, the full list has beaten the market more than 2X over with an average gain of +23.9% per year. So be sure to give these hand picked 7 your immediate attention. 

Want the latest recommendations from Zacks Investment Research? Today, you can download 7 Best Stocks for the Next 30 Days. Click to get this free report

This article originally published on Zacks Investment Research (zacks.com).

Zacks Investment Research