Nvidia: Should investors avoid the tech giant?

One of the dominant forces on the S&P 500 (^GSPC) is Nvidia (NVDA). The tech giant has seen a record-breaking rally, but should investors continue to buy at its height? Theory Ventures General Partner Tomasz Tunguz joins for the latest edition of Good Buy or Goodbye to give insight into how investors should play the tech sector.

Tunguz calls Microsoft (MSFT) a "good buy," citing a multi-billion dollar run rate business in AI, its leadership in enterprise AI, and ample room for growth given AI's limited penetration in Microsoft's customer base. However, Tunguz highlights Microsoft's fraying relationship with OpenAI as a key concern.

Tunguz says "goodbye" to Nvidia, citing the rise of competition in the space. He warns that the company may need to diversify into software, which will be an uphill battle for the company moving forward.

This post was written by Nicholas Jacobino

Video Transcript

It's a big, noisy universe of stocks out there.

Welcome to goodbye or goodbye.

Our goal.

To help cut through that noise to navigate the best moves for your portfolio.

And today we're looking at the most valuable public companies as they battle it out for the top market cap spot.

I'm here with Theory Ventures founder Thomas and interesting.

Here.

You're looking at these mega cap tech stocks and let's get to your buy stock.

That is Microsoft now.

Microsoft obviously has already had a big run, right?

It's helped by all of this optimism about a I, but you think it's still a good buy.

So let's get to why.

First of all, you think that the potential in a I is even bigger.

It's enormous.

Microsoft, if you look at it, is basically the leader in a I.

They've developed a 5 billion run rate business in about 18 months, and, uh, because of their relationship with Open A I at least initially and their significant investments in data centres spending about 15 billion a quarter on Capex, uh, just in the data centre alone and since we're just in the early days, really, I think the stock has a meaningful way to run.

If you look at their relative share compared to Amazon and Google, they're actually winning share away from the other clouds.

And then the last dynamic is five years ago.

Most start ups really wanted when they started to build their companies, they wanted to build on Amazon.

And today most companies are actually building on top of Microsoft because they want access to those A. I features early on top of azure specifically.

Yeah, interesting.

So, um and a I is really then where they sit and where the strength is for a Microsoft, right?

I mean, and obviously they have all those other relationships outside of Azure with all the other software that they provide as well.

That's right.

I mean, you look at Microsoft.

I mean by far the largest software company in the world.

The major motion that they're now pursuing is the ability to cross sell.

So if you're a software engineer and you want a I, Microsoft will sell it to you.

If you want a laptop with a I, Microsoft will sell it to you.

The other dynamic is they support many different A I families of models.

So there's the open A I models.

There's the Facebook meta models.

Like the llama models, they support all of them.

They're developing their own models, and as a result of that broad penetration within enterprise, they have access to lots of data that other people just don't have.

And so they can train and they can infer and build new products.

And that cross sell, I think, will really lead them to build a pretty significant a. I business.

I mean 5 billion in 18 months, growing 700% year over year.

It's pretty amazing ability to create market cap.

And even with that growth, you say there's still a lot of room to grow because the A i stuff that they are now providing.

Not all customers are using it yet or they're not using the full scope of the capabilities.

That's exactly right.

So large numbers of customers I think it's something like 80% of the Fortune 500 are now using a I, but the penetration is still really quite small if you look at all the different products.

Nadel Satya Nadella done just a phenomenal job of prioritising within each of the product lines a I being the the most important feature that they sell, and it's starting to gain significant penetration.

But we're still really in early days.

The impact of some of these businesses in terms of efficiency is absolutely massive.

So Microsoft themselves are seeing a 50% improvement in developer productivity through their A I service now 75% you saw Klarna announced a quarter ago that they cut two thirds of their customer support team as a result of a I.

So there's a lot of efficiency that Microsoft's A I in particular and their systems will bring to these very large companies and they'll be able to read disproportionate share.

So we always like to point out potential risks when we're talking about one of these goodbye, goodbye cases.

And in this case, it's maybe that as you mentioned, the Open A I partnership is now a strength.

But maybe it won't last forever.

It won't last forever.

When you look at what open a I announced two weeks ago with the integration into Apple, you can see that Open is looking to broaden out its distribution, and that's really important.

Open A and Microsoft had a strategic relationship.

Microsoft invested about 13 billion and, uh, open A was using a lot of the azure chips in order to train and run the inference.

Microsoft then decided to start to sell directly to their enterprises, competing a little bit with Open A I and then open A. I is now pushing more in a consumer search direction.

So I think you'll see open A and Microsoft start to compete more and more, which will challenge that pretty key relationship.

And there may be some headwinds there.

I think that risk is mitigated by the fact that there's so many model families, so many different kinds of a I that Facebook and others are creating that Microsoft should be able to differ certify.

But this is an important risk for sure.

Yeah, interested and keep an eye on.

All right, let's get to the stock that you're avoiding right now, and this is I gotta say, you might be in the minority on this one.

We're talking about NVIDIA because you know, most of the folks we talk to are big fans of NVIDIA, and they think it's gonna continue to grow.

Obviously, the stock performance has been incredible.

I mean, when it gained as much as it did last year, we said There's no way that can keep going And it has So But you you talk about the competition, which is something that is out there.

Certainly a lot of folks have brought up this issue, right?

So NVIDIA, unbelievable performance, right?

You've seen revenue increase.

Five X.

You've seen profitability go from profits go from about 4 billion to 4.

40,000,000,046 billion four years.

Really incredible business.

It's really driven by a I and the inference when you have those profit margins like 60% you're going to face a lot of competition.

And it's not like I mean, Google has been developing their own TPU chips.

Amazon has both training and inference chips.

A MD just announced the chip that is 30% more powerful than the current NVIDIA H 100 which is the dominant inference chip.

And so there will be significantly more competition here.

And if you look at NVIDIA's profitability growth, it's basically just a doubling or tripling in prices in terms of the GP US and as more competition comes in as the Capex spend increases pretty materially on the development of these chips.

I think those profit margins ultimately shrink.

Interesting.

OK, And let's also talk about valuation because that's a part of the story.

Um, so if you look at the PE that we have seen and this is the forward PE, this is going back, Um uh, five years, I believe on this forward PE and you have seen sort of spikes in the PE.

And then it's come back down, right?

So what is that?

I mean, we're not quite at the spike yet here and in fact, a lot of the people we talked to say, actually, the the valuation hasn't gone up that much.

But how are you thinking about the waves that we have seen?

So in video, if you look at the stock over a 15 year basis, has seen three major waves.

The first wave was around gaming in the 2010 2012 time frame.

The second wave was driven by crypto uh, in sort of the 2018, 2020 wave.

Yeah, and now we're sort of in the a I wave, and at each point in those waves, the PE has gone above 50 then 68 quarters later has fallen to 25.

So the counter argument would be, well, this wave will run a lot longer because, uh, the and the the NVIDIA team has said, at least through the end of next year, there's plenty of demand for GP US that outstrip supply.

But over time, the reality is like this is a company that's selling chips, and these kinds of multiples are, I think, really difficult to sustain over the long term, right?

It's quite unusual.

So they need to look into software, and they already have, right, they they're selling sort of the sweet, including the chips, including the whole, um, the rack as part of the whole selling point and the software.

But you think they need to get more.

I think they absolutely need to get more into software.

I mean, you think about, um, the multiples that software companies have.

They typically trade as a function of revenue, particularly the high growth ones, and I put in VD.

If you were to think about it as a software company, would probably trade it somewhere between 12 to 18 times top quartile, maybe even a little higher just because of the dynamics around a I and how special they are.

But the reality is the fraction of revenue today it's in software is is not disclosed, and it's probably still quite small.

So as the demand for GP US attenuates with time, they will need software to meaningfully boost up their multiple, particularly because the growth rates of the data centre business will plateau today.

You're seeing, uh, Microsoft and Google and, um, and Amazon spend about 12 to $15 billion in the data centre, a quarter over time.

That will sort of that will sort of decrease.

And so the growth rate in the data centre part of the business, which today is about 60% of the revenue and by far the fastest growing compared to gaming and automotive will slow.

And so software will absolutely be the key.

Very interesting.

OK, so let's get to what, though could go right for NVIDIA.

And that's just that, you know, there is this promise of competition, but it may not come for a while.

That's right.

So I I'm a long term investor, so I'm thinking about 3 to 5 years in terms of how people actually compete with NVIDIA, you need to produce chips, right?

And within the US.

Our ability to fabricate chips is pretty limited because we've outsourced that through the whole fabulous wave in the eighties and nineties.

And it's it will take a long time to get these fabs fabrication manufacturing facilities up and running.

And the time to ship some of these chips is a lot longer.

It's just in software, right?

There's tape out.

You have to actually produce it.

There's yield management.

There are all kinds of supply chains, and so it may take longer than and particularly.

I mean, you look at Facebook will spend alone $20 billion this year buying GP US, Amazon and Google who are building their own chips.

They may not be able to ramp production that quickly or even, uh, satisfying that demand.

In fact, Microsoft right in the last quarterly, uh, release said that, um, they're unable to satisfy the demand, so yeah, so OK, well, we'll see what happens with NVIDIA real quickly before we let you go.

What are your positions in each of these stocks?

So I hold, uh, Microsoft is my single, largest public position.

Uh, NVIDIA is a very, very small position.

Got you OK, Tomas.

Thanks so much.

Really appreciate it.

Interesting stuff and a little bit of contrarian.

Take on NVIDIA.

Appreciate it.

Thank you so much for watching goodbye or goodbye.

We'll bring you new episodes at 3:30 p.m. eastern.