Nvidia’s AI Chip Dominance: Can Custom Chips Threaten Its Valuation?

During their recent episode of the Acquirers Podcast, Travis and Carlisle discussed Nvidia’s AI Chip Dominance: Can Custom Chips Threaten Its Valuation?. Here’s an excerpt from the episode:

[Tim] It has. I mean, it’s a pretty different business now, but there you go, Toby. Look, all of the hyperscalers are building their own chips now.

So they’re using Broadcom and they’re designing their own chips that are different, but they utilise less energy. And so they’re more energy efficient. And I mean, that’s interesting.

So I guess if you look at Nvidia from a PE standpoint, it looks fairly cheap once again on the ETH part of it. But I mean, that’s a pretty challenging competitive threat out there. I would think long term.

I mean, these companies are developing very viable TPUs and chips. So Amazon, Google, Microsoft has them. I don’t know.

I believe Apple is doing it. I’m not sure on Apple.

[Tobias] Could be great for the consumer if all of that comes through because you’re competing AIs that are pretty close. I mean, that was the… I was in China last year, in April last year, and one of the things that we were told by the VCs there was that they viewed the American AI.

And this is 18 months ago now. So this is old technology, but they were clearly thinking about the future there as well. They could see that everything was leapfrogging.

At the time, the American version of open AI was ahead of everything else. But DeepSeek, which is the free Chinese version, was sort of just behind it. And I wonder, for most people, do they really need the bleeding edge, expensive AI to achieve most of the things that they want to get done?

Or can they do it with a cheaper version that’s just a generation or two behind, that gives you 95% or 90% of the performance? I think that’s probably the case. There’ll be people who just use an older version off the shelf, don’t pay much for it, does most of the stuff you need, and it’s the people who…

A handful of people need the really bleeding edge stuff, and they’re probably going to pay for it. Great for a consumer, hard for a business to make a whole lot of money with that. It’s like when the hard drive makers first started coming out, they just leapfrogged each other so fast, they’d be a winner for a little while, but something else would come through, like basically an uninvestable sector, unless you really, really knew what you were doing.

And I think it’s just easier to avoid that stuff. You’re a value guy, you don’t have to be playing in all of it. It does make it painful, though, when there’s a little bit of a mania going on and everybody’s in those businesses.

[Tim] Think of Yahoo. Yahoo was the top dog, right? And then Google started just eating them up.
Remember what Yahoo’s market share got down to? The company just could not perform. I don’t see this as being that different.

Let’s say the market starts picking winners. Let’s just say, okay, we’re going to write off one of these, and we’re going to focus on a few of these, and the Chinese cheaper versions or whatever. You’re not going to be able to raise capital at the same levels, you’re not going to be able to spend at the same level, so you could really fall behind the arms race.

It’s just a really challenging time. I don’t think that companies like Amazon or Microsoft can really afford to sit around and let Muse just gobble up the consumer part of it. I mean, Amazon’s not letting…

I mean, why wouldn’t they do that themselves? I can’t imagine that they wouldn’t. Unless it competes with their customers, I mean, that’s something to think about at AWS.

I mean, maybe like Anthropic and OpenAI, they spend a lot of money with Amazon, but who knows?