Nvidia, the company that dominates the graphics processor market, was once best known in the video game world. But these days, Nvidia GPUs are also the go-to choice for the massive processing power needed to run generative AI systems — and the recent explosive AI hype has catapulted the company’s stock into the stratosphere. Nvidia briefly peaked at a trillion-dollar valuation, aligning itself with tech giants like Alphabet and Apple, and sparking a kind of turmoil in the markets. Nvidia appears to be the first big stock gain of the AI era, and investors’ mouths are watering.
On Sunday’s episode of What Next: TBD, I spoke to Don Clark, a freelance reporter specializing in the chip industry, about how Nvidia drove the AI revolution, becoming the hottest chip maker in the world and transforming the entire AI industry. Made madness suddenly seem very real. Our conversation has been edited and abridged for clarity.
Emily Peck: In 1993, Nvidia was just an idea dreamed up by CEO Jensen Huang and a couple of his friends in a Denny’s – the stuff of the Silicon Valley mythos. What were you concentrating on at that moment?
Don Clark: They basically made it their mission to do the graphic work. Gaming was a big deal. CPUs didn’t do very well, so they started out in this business and made it really tough – basically everything computer-related, they’ve been there. They’ve always been involved with the chips that make movies, chips that scientists use to visualize things. When it comes to everything to do with painting pictures on a screen, Nvidia has constantly pushed the boundaries. Since the company was founded, they have invested $37 billion in research.
Is there something about Jensen that puts him at the forefront of this matter, something that makes him take risks that others don’t?
In general, those are the glory days of any business when your CEO is very tech-savvy. When Bill Gates ran Microsoft, they kicked ass; When Steve Jobs ran Apple, they were awesome. No engineer can say to the great leaders who have really helped the industry, “Oh, that’s too complicated” or “We can’t do that.” Jensen is very, very technically savvy, he reads all the scientific literature and speaks directly the scientists. A young misfit in 1993, he’s 60 now – but he still exudes that energy.
Nvidia is excellently positioned to capitalize on the AI onslaught. Their chips are some of the most advanced on the market, and they’ve been the leading chipmaker powering AI systems for nearly a decade. What does this mean for the company?
It can be argued that AI will change every bit of software. Think of every single thing that has intelligence in it, from your toaster oven to your car—everything has software in it. Jensen is basically describing how the many trillions of dollars of investment in the world’s data centers are all being transitioned to this way of working with AI. Nvidia was notoriously involved in phones and decided to get out of it. They just decided to focus on data centers.
This isn’t the first time Nvidia has seen tremendous growth due to a new technology powering it. During the crypto boom, Nvidia GPUs were in high demand among crypto miners, and as the crypto industry became a breakthrough, Nvidia stocks soared only to come back down to earth along with the crypto markets . What characterizes the AI boom?
Essentially, crypto miners were competing with gamers for the same type of GPUs that video games use, and that’s how the stock bubbled. It was very difficult to quantify as no one knew exactly how many of the Nvidia numbers reflected the crypto boom, and so they were hit as things sort of unwound. But the AI boom is about transforming everything we do. I mean, every Google search you do now is powered by AI; Every time you’re looking for someone on Facebook and people are recommended, that’s AI. It’s not a fad, it’s inevitable.
I’ve been watching the markets for Axios, and we’ve watched Nvidia stock do really well this year. But when the company recently hit a trillion-dollar valuation — along with Apple, Amazon, and Microsoft — it felt like it came out of nowhere. Did it surprise you?
I knew it was going to be a big quarter, but I didn’t expect it to be like this. Markets are always six months ahead of the prediction of these cycles. Last year was a terrible year for semiconductor stocks. But if you think the cycle will turn, you want to be the first to bet on getting there, so basically there have been a lot of hasty bets that that turn will happen.
Can the US Bring Chip Manufacturing Home?
Nvidia is experiencing a rush of orders for its chips – why is that? Was it the launch of ChatGPT that increased competition across the AI market and pushed companies to expand?
That’s basically what happened. Microsoft and OpenAI struck this deal and it was announced last year, with Nvidia stating that tens of thousands of GPUs were involved. So everyone else realized that they needed to step up their Generative AI game. All of these cloud providers basically use GPUs in two ways. They have their own AI jobs that they do internally for themselves and then, with the exception of meta, they also sell compute cycles to other people. So if you’re a startup and want to run AI, you usually don’t buy a GPU, you go to AWS and use their Nvidia-based computing service.
But because companies rely so much on Nvidia for their infrastructure, Nvidia’s biggest competitors are also their biggest customers, right?
I’d say Nvidia’s closest competition is Google and Amazon’s in-house developed AI chips, and because they know the job pretty well and can tailor their software to it, they can get a pretty good cost reduction in doing that AI work .
Google is best known for this. They do some of their own work and sell some work on their own TPU chips. Amazon, which has developed many of its own chips for other parts of the computing industry, recently launched some AI chips of its own that it says are slowly catching on, albeit in their infancy.
Temple Grandin: AI will not destroy us – if we make one crucial change now. Apple is ignoring something big about augmented reality. The true risk of wildfire smoke for children. This won’t be the last time the East Coast experiences wildfire smoke. Here’s how to deal with it.
I believe you everything is going to be AI based and that’s the future – but it still has that feeling when you look at the stock market, Nvidia’s stock price and all of the recent coverage… it’s got a bubble vibe. It reminds me of the dot-com boom.
I’m not defending Nvidia’s current stock price as there is a lot of speculative stuff in there. And the money is flowing like, “Oh, I’m a fund and I need some AI in my portfolio, who can I buy?” So if we had this conversation two years from now, I would think Nvidia’s stock price would be there would be where he is now? I wouldn’t say it’s guaranteed at all.
I would not disregard the revenue they will generate with AI. The question will be how that affects how inflated the stock price is already. It’s one of those things about the entire semiconductor industry: it’s not enough to build one great chip, you have to have a plan that’s going to get better, better, better, and each and every one of those things costs hundreds of millions of dollars to develop. So it’s not a game for the faint of heart.
How has Nvidia made it easy to get into the AI business?
People underestimate the software. Nvidia makes it relatively easy to get into the AI. They have a university where you teach people how to make things, they have their own supercomputers that sell bikes to other people, they sell reference designs for other server companies to make things, they have all these pre-trained AI models that they provide you with for specific jobs and specific industries. They have built a huge software moat around their company.
Future Tense is a partnership of Slate, New America and Arizona State University that explores emerging technologies, public policy and society.