expected to begin in the first half of 2028, and it comes at a time that the U.S. is in desperate need of more memory chips to meet the massive demand for hyperscale and AI enterprise customers. And that is among the top reasons Nvidia officially announced increasing prices on the upcoming Grace Blackwell GPU chips. The semiconductor giant reported a second-quarter top and bottom line last night, and revenue exploded 97% year-over-year to a record $96.2 billion, soundly beating the estimate of $92 million—soundly—and the 15th consecutive quarter beating expectations. And after that report, I spoke with CEO Jensen Huang for the impending price hikes and bottlenecks to keep up with the epic demand. And Jensen, I do not see a single indication you or any of the AI players who deal with it are stumbling or taking a breather in this AI race. And if you had to use one word to describe another recent quarter and the rest of the year coming up, what word would you use?
Demand accelerating. Not only is demand already strong, accelerating is really quite extraordinary. The reason for that is this: reasoning has become useful, productive work and is doing productive tasking. And the tokens that these AI labs are generating are profitable tokens. And when you're generating something profitable, the demand is so extraordinary, companies want to generate more of it. So the thing that's limiting the growth is not customer technology. The customers are—the demand is incredible. What's limiting right now is just compute. We need to bring in more compute online so that we can help these AI labs grow.
And the second thing, and this is a part that people do not see: AI is way beyond more than just a few top AI labs. They are doing incredible work there, but AI is literally everywhere. It is being used in enterprises and used all around the world, neo-clouds and sovereign clouds. And the other half of the business that most people do not see, that is also growing incredibly. So we see growth everywhere. And next year, launching a new product, Rubin, and the success of it is incredible. This will be the fastest product launch in history. What was incredibly awesome with Blackwell, Rubin will be even faster. So it's going to be a huge year next year.
You predicted that Vera Rubin would be best. Can you say that every single Vera Rubin chip that comes out of the fabrication plant is spoken for, and how far out do the orders go?
Customers have to plan with us at least a year out these days because these are not buying a computer; they're really building a factory. And these factories are $50 and $60 billion in investment, and so you are planning pretty far down the road. And we have complete visibility into our demand through all of next year and beyond. But for next year, every single chip that we can make has already been sold. So we are predicting at least 70%—supply constraint, and our demand is extraordinary.
Speaking of looking at your unit, said next year believe revenue opportunity could, I guess, exceed $1 trillion, and that would be double the 500 billion that you're expecting this year total. So I'm just wondering, with such massive demand that you just articulated, is that $1 trillion number already old news and is on the move, could be higher?
We are forecasting 70% growth year-over-year, and so not quite trillion yet, but the demand is greater than 70%. And it's just that we only have supply chain support for 70%. And working with the suppliers—we have the largest supply chain in the world, we have great partners—but the demand is just too strong. And so we need to make sure they continuously work on the supply; otherwise, the customers will be disappointed. The demand is higher than the supply we currently have.
Can you pick up the phone with your big manufacturer and tell the guys to go any faster and make more?
It's TSMC packaging, wafer-level packaging, and for memory and everything—networking, silicon photonics, and power generation in the data centers. It's availability of high-quality land that has power and is ready to receive these computers. And, you know, you have bottlenecks literally everywhere because this is the largest infrastructure build-out in human history. Creating jobs in chip plants, packaging plants, computer plants, and, of course, all of these AI data centers. And creating jobs everywhere is also a challenge because there's just not enough skilled labor around the world and certainly not here in the United States to support our entire demand. We have enough support for the 70% that we forecasted.
Bottlenecks—because the memory chip bottleneck is an issue, and one of the big questions that have been hanging out there was actually answered in the conference call by your CFO, who said that, come the first quarter of next year, the price of your chips will go up. Rumor has been it would be up about 15%. What are you telling the customers it will be, and is that around the correct number?
This is one of the concerns that most investors have had, and they have seen that memory prices have gone way up. So this quarter, we decided that we would rip the Band-Aid off and reset expectations about the gross margins, and just explain to people that, in fact, we've now absorbed the cost increases and we have also repriced our products in the marketplace. And based on that, our margins will come down from 75%, but it will be between 72% and 73%. We have now reset the expectations. And, you know, we removed this anxiety that everybody had about gross margins, and it's kind of been a bit of a cloud over our stock for some time. And I'm just happy we are able to let everybody know we now have everything worked out and we rolled out new pricing. We've accepted our cost increases, and everything from this point forward, let's go execute.
