Can NVIDIA Become Intel Faster Than Everyone Becomes NVIDIA?

In 2013, then-Chief Content Officer of Netflix Ted Sarandos quipped that the company's goal was "to become HBO faster than HBO can become us." In hindsight, the quote is sort of silly – and naturally, Sarandos regrets it – because in 2024, Netflix is far larger than HBO ever was, with an entirely different scope. Netflix isn't just HBO, it's HBO plus about 100 other channels. It's closer to all of cable itself than it is to any one channel.

Anyway, it's still a great line because it succinctly framed Netflix's ambition as they moved into original content. And it framed the problem HBO faced in the new world of content which Netflix was shaping. But the landscape was ultimately shifting too fast for anyone, including Netflix itself, to know how it would ultimately shake out.¹

Today, no industry is moving faster than AI. And while OpenAI may be the new company most associated with the cycle, no company is benefitting from it more than NVIDIA. What started as a startup making graphics cards for PCs is now the third most valuable company in the world. And it may become the second soon, depending on how Apple's AI message at WWDC in a few weeks lands with Wall Street. Like the current most valuable company, Microsoft, NVIDIA has fully embraced the AI narrative. But unlike Microsoft, AI is now the dominant business for NVIDIA. Which leads to the inevitable question: can they hang on?

Even those most bullish about AI right now will readily admit that we're likely in a new bubble-forming part of the cycle. Blame it on investors, who have a lot of capital and few places to park it, certainly when it comes to places with the potential to lead to "grand slam" outcomes. "Web3" was a tease, AI is the real deal, the thinking goes. And it's not necessarily wrong at the high level, but on the ground floor, most of this capital is going to get washed out as the industry shapes up and shakes out. That's fine, that investment is accelerating everything and no one will be sad when much of that capital isn't returned, let alone multiplied – beyond some VCs and their LPs.

But unlike with the crypto craze, there are also a number of public companies benefitting from the AI boom cycle. And when that cycle ends, it will likely spell trouble for many as the stock market as a whole is so heavily weighted by the large tech companies. All of them will be able to weather any such correction, of course. But NVIDIA is the most vulnerable because again, so much of its business and future story is based around AI. And it's a good story to be based around! But the market is ballooning the stock up as the one bet you can easily make in "pure" AI at the moment. And again, regardless of the future of this technology, it's going to correct at some point, as it always does.

At the same time, basically every company with any sort of chip-making and/or designing capability is coming after NVIDIA. None can hold a candle right now, with even the potentially best-positioned rival, AMD, lagging behind. But all of these companies from Intel to Google to Apple to Qualcomm to Microsoft, not to mention a number of startups, are working to try to catch up and/or outmaneuver NVIDIA on some other angle of AI.

NVIDIA of course knows this. Founder Jensen Huang has built the company into what it has become by making prescient bets and maneuvering to be in the right place at the right time. Even after the crypto run-up subsided, and NVIDIA looked deflated, the company found itself perfectly positioned to take on the AI task, quite literally. And so what can the company do now to insulate and prepare itself?

Well, it sure looks like the answer is to speed up. Huang surprised many this weekend by announcing the follow-up chips to their just announced-in-March "Blackwell" chips. The company had said they were going to upgrade their chips each year, and they weren't lying. But the pre-announcement of "Rubin" is aggressive. And it would seem to be a signal to those would-be rivals and, in turn, to Wall Street: Bring. It. On.

At the same time, NVIDIA is moving further into tangential fields with not only different types of GPUs, but also CPUs as well. At first, it seems the aim is to try to take on more of the datacenter workloads, which their rivals Intel and AMD have long controlled. But presumably, this portends a move to try to take on all computing chip needs in general.

All of that is to say, NVIDIA is trying to become Intel faster than Intel (and everyone else) can become NVIDIA.

It's sort of a quip, but also not. While NVIDIA is now many times larger than Intel ever was in terms of market cap,² Intel was once synonymous with computer chips thanks to the rise of "Wintel" PCs.³ Qualcomm is clearly trying to make a move to be that player tethered to Microsoft with the new "Copilot+ PCs" (doesn't roll off the tongue nearly as well). While AMD and yes, Intel, will soon make their own cases to be the Intel of this equation. And NVIDIA is inserting themselves here as well – not with PC CPUs, at least not yet, but instead partnering with AMD to overpower, quite literally, the rise of NPUs being pushed by CPU makers.

Microsoft made a big to-do about NPUs in the recent Copilot+ PC press cycle. As did Apple before them in the unveil of their M4 processors. Google has their TPUs, of course. And everyone (aside from NVIDIA) wants whatever the new version of Moore's Law ends up being to push this type of computing to parity, faster.⁴

How will it all shake out? Probably similar to the way Netflix vs. HBO did. Which is to say, completely different than the way anyone thought. Again, I go back to the notion that AI itself is moving and shifting so quickly that any number of breakthroughs could change the game overnight. That's not hyperbole, it just might happen at any moment. And you could certainly argue that the companies actually making and training the LLMs have an advantage in creating new types of chips perfectly suited for their tasks. Unless and until LLMs matter less than whatever comes next. Which seems like a good bet. But these are all insanely expensive bets to make on technology that is still so unpredictable.

And so NVIDIA will continue to benefit for quite some time, you have to imagine. But they need to use this time to lay the groundwork for what's next. A decade ago, that general stance, more so than the "HBO" target led Netflix to both create and take over streaming, eventually luring their competitors into overspending to try to catch up, not realizing the market would shift and they could not. Will it play out the same with NVIDIA? Or completely different?

I asked ChatGPT to predict the market cap of NVIDIA in 5 years. The guess? $2.49T, which is, of course, lower than it is right now. An interesting guess. Sadly, based on a false bit of data AI pulled in assuming the market cap right now is around $1T. (And also a far lower revenue growth rate than the company has been seeing of late – though noting it is likely to fall from unsustainable heights at some point.) So, it either inadvertently believes NVIDIA will fall a bit with a change to the market, or will grow 2.5x from where it is now. Which would be... $7T. We'll check back.

¹ Though I was pretty close back in 2011...

² Intel's peak market cap was just over $500B in 2000. NVIDIA is quickly approaching $3T. The company has gained over $1.5T in market cap just in the first half of this year. I'll do the math for you: that's 3x Intel's peak market cap. In six months. Today Intel's market cap is just $130B.

³ Though Intel didn't hit their peak revenue number until 2021, at just under $80B. NVIDIA's top annual revenue was last year at just over $60B. But that was on insane 125% yearly growth. Oh and last quarter, NVIDIA's $26B in revenue pushed them past Intel's peak if you look at the previous 12 months of revenue. And last quarter's number grew at 262%. 🤯

⁴ NVIDIA has a software moat as well with CUDA, but that's also already under assault -- pushed by who else? OpenAI.