27.08.2026 - Nvidia: Crushes the Numbers

Nvidia delivered another massive quarter last night, reporting USD 96.2 billion in revenue, while Q3 guidance came in at an extraordinary USD 108 billion.

Even more impressive was the longer-term outlook. The company indicated estimated revenue growth of around 70% for fiscal 2028, significantly above previous Wall Street expectations.
Based on the current consensus projection of approximately USD 396 billion in revenue for fiscal 2027, another 70% increase would take Nvidia's annual sales to roughly USD 673 billion.
At that level, Nvidia would overtake Apple and Alphabet based on current Wall Street revenue projections and rank behind only Amazon among the largest US technology companies.

And according to CEO Jensen Huang, even that extraordinary growth rate is constrained by supply rather than demand.

“Our demand is much greater than 70%,” Huang said during the earnings call. “Our supply allows us to confidently deliver 70%, and we’re going to continue to work with our supply chain to increase on that.”

Markets: Nvidia shares are up more than 7% in pre-market trading

• Equities: AI-related stocks are rallying, with Nasdaq futures up more than 1%.
  

My View: The strong quarter itself is not much of a surprise. Neither is the strong outlook at current stage.
Nvidia remains at the very center of the unprecedented global AI infrastructure buildout, and Jensen Huang's comments make one thing very clear: the immediate problem is not demand. It is supply.

But this is exactly where I continue to question the sustainability of the broader AI cycle. Why?
Because in my view, the AI sector is increasingly driven by the same force currently dominating parts of the equity market: FOMO.

Companies are afraid of falling behind and potentially losing the AI race. As a result, they are buying as much computing capacity as they can secure. The primary question currently does not appear to be: What return will we generate on this investment?

Instead, it is: What happens if our competitors invest and we don't?

That creates an extraordinary demand dynamic. But extraordinary demand today does not automatically mean extraordinary returns tomorrow.

Remember the Toilet Paper?
Think back to the beginning of the pandemic. There was never fundamentally a shortage of toilet paper. But people became afraid that there could be one. That fear itself created the shortage. People rushed to stores and bought far more than they actually needed because everyone feared being the one left without any.

I increasingly see similarities in today's AI chip market. Companies fear being left behind. Therefore, every available chip is being bought. The enormous demand then reinforces the perception that even more capacity is needed, encouraging companies to invest even faster.

The difference, of course, is that we are not talking about toilet paper. We are talking about hundreds of billions, potentially trillions, of dollars in capital expenditure. And that capital is not free.

The bigger question: Who ultimately pays for It?
Nvidia's numbers demonstrate how much money is currently flowing into AI infrastructure. They do not yet answer the much more important long-term question: How much money will ultimately come out of it?
The hyperscalers are spending enormous amounts on chips, data centers, energy infrastructure and networks. Increasingly, part of that expansion is also being financed through debt.

At the same time, several constraints are becoming increasingly difficult to ignore:
Higher bond yields make financing these investments more expensive.

Data centers require enormous amounts of electricity, while power availability and grid capacity are becoming bottlenecks in several regions.

Operating costs remain substantial, even after the infrastructure has been built.

Resistance against new data centers is increasing in some communities because of electricity consumption, water usage, land requirements, noise and infrastructure pressure.

And perhaps most importantly, the end consumer is showing increasing signs of weakness.

That matters because somewhere at the end of the AI investment chain, someone eventually has to generate enough additional revenue and cash flow to justify these enormous investments.

At the same time, pressure on consumer credit is increasing, banks are becoming more cautious and tighter lending standards could further constrain economic activity.

None of this questions Nvidia's current operational strength. The company is delivering extraordinary numbers and currently sits in perhaps the strongest position anywhere in the AI ecosystem.

Another important driver behind Nvidia’s extraordinary revenue growth is its enormous pricing power. With demand exceeding supply, Nvidia can sell not only more chips, but also increasingly expensive chips. However, this raises an important question: How sustainable is that pricing power?
Therefore, today’s exceptional revenue growth should not simply be extrapolated into the future. More competition could eventually pressure both Nvidia’s pricing power and margins.

Nvidia's success today does not automatically validate the economics of every dollar being invested across the AI ecosystem.
That distinction is becoming increasingly important. For the current growth trajectory to continue at anything close to today's pace, AI ultimately needs to generate enough productivity gains, revenues and cash flows to justify hundreds of billions in infrastructure investment.

Maybe it will. But today's valuations already assume that it will.
They leave very little room for a scenario in which AI infrastructure spending slows, financing costs remain elevated, power constraints intensify or companies simply begin asking a question that currently seems secondary:

What is the actual return on all this investment?

Nvidia just crushed the numbers. The bigger test for the AI cycle will come when its customers have to prove that they can do the same.

Become a member to access more valuable market updates like this