Let's talk about Nvidia earnings.

It was three years ago, almost to the day, that Nvidia reported a growth shock that changed the world.

Remember, Jensen Huang told us that the world was beginning a trillion-dollar transition from general purpose computing to accelerated computing. And he said it started with the ChatGPT moment (the launch of OpenAI's generative AI model, which crystallized productization of AI). 

He said the demand for Nvidia's AI data center chips was so steep that the growth he just reported (in May of 2023) would be even bigger the following quarter. Moreover, he said the hyper-growth would continue for the foreseeable future.

That was the Nvidia moment. It was the moment the world realized AI was about to reinvent computing.

With that, the transformation has been bigger and faster than Jensen predicted. Still, there has been plenty of doubt about the significance and the durability of the AI revolution along the way. 

Then we had a series of "moments" earlier this year, that gave clear signals that the entire AI ecosystem wasn't just sustaining, it was removing the ceiling on what is thought as possible for the economy.  

First, it was data storage. Sandisk reported $3 billion in revenue and then guided to $4.4 to $4.8 billion for the very next quarter — a 50% sequential leap in 90 days. The same shape as Nvidia's famous 2023 guide. It was the moment the world realized AI doesn't just need chips. It needs endless storage to hold the oceans of data those chips produce.

The Nvidia moment for storage had arrived. 

Then it was the chips themselves.

Taiwan Semiconductor, the company that manufactures the world's most advanced semiconductors, reported a 20% jump in revenue in a single month, in what is normally a quiet, post-holiday lull. At the same time, its board greenlit a $45 billion plan to build more capacity. The bottleneck was starting to break.

The Nvidia moment for chips had arrived.

Then (in his Februrary earnings call) Jensen named the next moment himself. He said the "ChatGPT moment of agentic AI" had arrived.

Consider what that phrase means.

Not the Nvidia moment, but the ChatGPT moment (like, the beginning!). 

"Agentic" AI is moving the AI revolution into another gear. It's the difference between AI as a tool a person uses, and AI as a worker that operates on its own.

And it's always on, and always inferencing (always reasoning, producing output).

This is why in his February earnings call, Jensen said this line over and over: "compute equals revenue."

Today, he said "it is very clear compute is revenues."  Then he added, "compute is profit."

Nvidia is proving that. 

The numbers in Q1 were jaw-droppers. Almost $82 billion of revenue in the quarter. That's more than 10x the revenue of just three years ago. Almost $38 billion of revenue growth from the same period just last year. This, with net income margins in the mid 50s(%). 

Then there's the guide. Nvidia told the market to expect $91 billion next quarter — another double-digit jump.

And Nvidia's CFO told analysts that AI infrastructure spending is now on track to reach three to four trillion dollars per year by the end of this decade. 

That's three to four trillion dollars, annually.

We've watched this estimate climb in real time. Last October, Jensen said there was half a trillion dollars of demand visible. By March, he had raised it to a trillion. Tonight, they put it at a three-to-four-trillion annual run rate.

With that, as we've discussed here in my notes, the day-to-day noise ( the rate guessing, the tariff headlines, the market swings on a given afternoon) is a conversation about ripples on the surface.

Underneath the surface, the largest industrial buildout in modern history is funded, profitable, and accelerating.