Is Nvidia really worth $4tn?
Nvidia is now the most valuable company in America, as AI seems to power an unstoppable market. Today on the show, Katie Martin and Aiden Reiter ask if it reminds anyone of the dotcom bubble. Also they go long snow globes and long stone fruit.
Sign up for the FT Weekend Festival at ft.com/festival and use the promo code “FTPodcasts” for 10 per cent off.
For a free 30-day trial to the Unhedged newsletter go to: https://www.ft.com/unhedgedoffer.
You can email Robert Armstrong and Katie Martin at unhedged@ft.com.
Read a transcript of this episode on FT.com
Hosted on Acast. See acast.com/privacy for more information.
Listen and follow along
Transcript
If bonds are back today, why wait for tomorrow?
At PGM, our fixed income strategies help investors uncover hidden value and unlock opportunities.
Whether you're looking to enhance your income or diversify your portfolio, our broad range of strategies bring together local expertise and deep credit research to help you achieve your long-term goals.
PGM, our investments shape tomorrow, today.
Pushkin.
Here's a company we haven't chatted about here for a little while, NVIDIA.
It's a chip maker.
It's basically the artificial intelligence trade, and it's now worth, wait for it, a little over $4 trillion.
That's trillion with a T, folks.
And for reference, the entire FTSE 100 index in the UK is under $3 trillion.
What the hell?
So, it looks like the whole AI hoopla is alive and well in financial markets, that's for sure.
So, why is it that whenever I see something that's AI generated, it's rubbish?
Today on the show, we're asking, is this a lot of money riding on a wing and a prayer, or is this technology the real deal?
This is Unhedged, the Markets and Finance podcast from the Financial Times Times and Pushkin.
I'm Katie Martin, a markets columnist here at FT Towers in London, and I'm joined by one of the youngsters on the Unhedged newsletter who makes Rob look clever, Aidan Writer.
Hello.
Now Aidan, one of the challenges with talking about AI is that like lots of normal people don't really know what it is and we don't really understand how the technology works or what it's going to do.
So a very confusing thing for us to be talking about today, but we're going to try.
Yeah, we're just going to not know, but we'll keep on going.
You've got to love that.
This is the OnHedge podcast way.
So let's start with a little bit more on NVIDIA.
I mean, like two years ago, it was worth a measly $1 trillion.
Who cares about that?
But it's just like so,
like, it's just unholy how massive this thing is now.
The shares are up about 40% since early May.
Like, why has, why has it just rocketed like this lately?
I mean, it's interesting that it's rocketed like this given that just, what, six months ago, we weren't sure it had any future.
After DeepSeek came out and proved that you can make these complicated AI models with the lower-grade chips, so not the chips that NVIDIA says will build the future, its stock plunged, I think, 17% in a day.
Or, you know, at least very, in the teens.
And then, of course, within two days, it was back up because it was very clear to me.
This is when China came out with its own version, effectively.
A Chinese company, DeepSeek, created a model using lower-grade chips.
So the chips that are not the fancy dude ads that NVIDIA has hawked saying this is the future of AI.
And they did it for, you know, relatively cheap.
There's a lot of debate on whether or not the numbers they gave are actually accurate.
But the point is that a lot of people doubted whether or not you really needed such advanced chips in order to make complicated models that do the work that AI is supposed to do.
But the market has clearly forgotten that because despite that concern, all the big tech companies that are buying NVIDIA's chips and plowing them into data centers have continued to do so and have done so at a very rapid rate.
Aaron Powell, I think there's also a sense that back in January, AI stocks were so richly valued, right?
They were so expensive.
They'd had such a great run that it only took a little bit of kind of bad-ish news in the form of a new Chinese competitor that the shares took a really big hit.
And then, of course, a little bit later on, we had the big shakeout in US markets that came from Trump's massive Liberation Day tariffs.
And so everything was just terrible for a little while in AI and in big tech for a few weeks there.
So,
you know, again,
it's just sprung back incredibly quickly.
And, you know, this is about numbers from NVIDIA, right?
This isn't just kind of, oh, you know, trust us, we're massive and important.
The earnings back it up.
Yeah, I mean, the earnings are incredible.
They've successfully sold a lot of chips to these data centers because people really believe that AI is the future.
The companies that have benefited so much from the quote-unquote AI hype continue to hype up the product.
And at the same time, as you were saying before, the market is continuing to buy it and continuing to ride the momentum, but people are jittery.
They're expensive stocks.
I mean, their PE ratios are very, very high.
That's the price-earnings ratios, which is how much investors are prepared to pay for future earnings that they think the company is going to make.
But let me tell you someone who is convinced that NVIDIA is the real deal, and that's its chief executive, Jensen Wang.
Now, I can't remember if I've mentioned this on the show before, but like at a certain point last year, I was on a panel session with a couple of people from the Irish tech scene.
And they were saying, we remember when Jensen Wang was just like some guy who would turn up to the same tech conferences that we did, and he was such a nerd and so unimportant that he didn't even get invited to the best parties.
And now, you know, they're doing panel sessions with me in a pub in Ireland and Jensen Wang is running a company that's worth $4 trillion.
And just recently, Wang has been going around Washington, D.C., talking to Trump and also going around Beijing.
And he's been talking about how AI is fundamental like electricity.
I mean, yes, but he's not the only one to think that AI is going to be transformative.
I mean, there's evidence to suggest it already is.
You know, we've had a long history of automation and new technologies in the world that have changed how we work, what we do, how we interact with one another.
AI seems poised to be the next one.
I mean, if you think about what mechanization and automation did for workers in the 19th century and 20th century, that completely changed how people were employed, what jobs were there in the economy, how people made money.
And there was definitely displacements, but by and large, it helped make the world a more rich and prosperous place.
Sam Altman at OpenAI says we're about to enter into the greatest period of human history.
Everybody's going to be more prosperous, et cetera.
And that's because what it's best poised to do is to replace knowledge jobs and knowledge work.
Now, some people argue that will be displacement.
Some people argue that will be enhancement, and we just don't know.
It's also probably going to affect physical tasks and physical jobs, like warehouse jobs, truck driving, and other driving jobs.
I mean, the impacts will not just be on the knowledge economy.
There's a lot we don't know, isn't there?
And I think we should just sort of set to one one side the argument that you hear some kind of excitable tech bros making, which is that AI is poised to become effectively sentient and it will take over the world and destroy all humans.
I'm willing to set that to one side as a tail risk.
So, you know,
whether you love this technology or hate it, it is the dominant force in the US stock market at the moment, which means it's the dominant force in global stock markets at the moment.
Like you can't ignore this stuff no matter how hard you try.
So a little note that caught my eye recently came from Torsten Slock, who's the chief economist at Apollo, the private equity group.
And he was talking about how the AI bubble of today is greater than the IT bubble of the 1990s.
And what I found interesting there was just this casual like dropping in of the B word, like this is a bubble.
It's like, oh, oh, okay, I guess that's one way to think about it.
And there is a kind of long-running argument about this.
But what we're certainly seeing is that big sort of AI adjacent stocks are getting all sorts of expensive now.
And they're more expensive than kind of shiny IT like tech stocks were at the back end of the 90s and that all ended badly.
So what's your sense of how people are trading this or how they should be trading it?
Yeah, well, I'll say first of all, I mean, I don't think the risks of an AI bubble are as big as the internet dot-com bubble of the 2000s.
For the most part, a lot of these like little companies that just attach AI AI to a very mundane product are VC-backed and in private markets.
It's not going to result in a big stock market crash.
Also, I mean, yes, there's a huge, huge hype around the seven big tech companies that make AI, but those are companies that have other revenues and other things they do.
So, you know, there is a floor, which is good.
But in terms of the whole market, I mean, if you look at NVIDIA's expensiveness, right?
It's price to earnings, what investors will pay versus what actually earns in money, that started ticking up
sometime in mid to late 2023.
Around the same time, you saw the same expensiveness show up in the rest of the SP 500.
So stocks got expensive right next to NVIDIA.
And that could just be like a dragging effect, but it also could just be, and this is sorry, stocks excluding NVIDIA and the other quote-unquote magnificent seven-tech stocks.
So the rest of the S ⁇ P 500 got expensive.
And some people will call that the AI halo.
And if you really believe that AI will pan out, there's reason to believe that those companies will benefit too.
The point is you can actually have some cost cutting across all these companies.
So you're saving on their bottom lines.
You think that they'll still have the same amount of revenue, but with fewer people.
So that inherently just makes them more expensive and increases their future earnings.
Aaron Powell, yeah, so I guess the question is, did other stocks rise at the same time as NVIDIA really hit the ascendancy purely because they were just sort of carried along for the ride?
Or is there a sort of sensible narrative that actually this really is going to transform the bottom line of corporate America writ large, like every single sector.
And again, my kind of problem with the latter argument there is always that my day-to-day interaction with AI makes me think it's like really rubbish, just like really poor quality.
It gives me like terrible search results that I can't get rid of at the top of my Google page.
It spits out faulty information, it spits out pictures of people with like six fingers and three legs.
And, you know, it just makes you think,
how is this worth so many trillions of dollars?
This just seems silly.
Well, so it's going to get better, right?
That's the common argument you hear from everybody.
All technology gets better over time.
It's the more mundane, number-based tasks like financial analysts or
just sending basic emails or writing very basic code that it's already shown to be very, very effective.
So I have a lot of friends who are coders and they say they use AI every day in their jobs.
I have a lot of friends who work in finance who say they use it absolutely every day, and it makes their lives easier because it's taking away some of the very basic numeracy and coding they have to do.
So, and it's not just going to be those people.
Imagine HR, IT, other people replaced by cheaper AI.
You don't necessarily need to find the guy in the office who knows how to use your computer if you can ask the AI how to use the computer better.
It's not better.
I'm not saying that's a world I want, but it's one you could very easily imagine.
Aaron Powell, yeah, yeah.
I mean, one of the things that people people often say about the dominance of AI in markets right now is that
the eventual winners from AI are not always
the companies that kind of originated it or that provide the foundational technology to it, like, for example, NVIDIA now.
It could be that 10 years from now, it's healthcare companies and it's consumer companies that are really kind of showing the benefits of this.
And actually, in 10 years' time, we might have forgotten who NVIDIA are.
Do you think that's possible?
I think that's definitely possible.
I mean, I'll say first on who benefits as opposed to the big tech companies.
We said, you know, there's cost reduction across the board.
There's also the chance for revenue enhancement.
So if you have each worker becoming meaningfully or marginally more productive, then you can have higher output.
And as long as there's people to buy that output, then you have higher earnings, right?
So it's, you know, the healthcare companies who can now see more patients or, you know, provide more at-home care care and be able to charge their insurance more efficiently.
It could be law firms or consultancies who can now
serve more clients and serve more people much more quickly and efficiently with the same amount of people.
Banks, that kind of thing.
Banks, yeah.
So you have revenue enhancement on top of cost reduction.
Again, this is assuming a scenario where AI doesn't put everybody out of work and people still have money to spend on things.
Yes.
That would be helpful.
Yeah.
So
that's the AI scenario where it helps everybody else.
And in that scenario, yeah, you could have big tech being a winner, especially if they're making the AI models that everybody's using.
But, you know, there's going to be winners and losers among those companies.
Some companies will opt for certain models.
Some models might prove to be better at certain things than others.
Some companies might not actually be able to properly scale and meet the specific and industry-specific domain demand in that world.
So it could be some of the big tech companies are huge winners.
There is also very high odds that they're losers too.
Also, they're most likely to come under whatever regulation eventually comes out around this, right?
About how they make their money, how they sell their models.
It's going to be much more burdensome for them to continue to be winners in the future.
Now, a lot of people who listen to this show are like not market specialists, and they probably do
look at what's happening with US tech stocks that are in this kind of AI universe.
And the first thing they will be thinking is they will hark back to that kind of dot-com boom and bust of like 20 odd years ago.
Do you think that is a reasonable concern?
I guess I'm asking, like, is this a bubble or not?
I mean, it sort of goes back to what I said before.
I mean, it's not as risky a bubble as in the 90s, from what I can tell, because not all these AI companies are publicly listed.
And also, it seems less risky because, you know, we've already started to see some real benefits and real impacts on the AI space and the ecosystem.
As we talked about, you could have cost reduction, revenue enhancement across a variety of industries.
What does feel more dangerous is how much the market is reliant upon these seven big tech companies that make AI, who may not be the winners at the end of the day.
I mean, have you heard anybody effectively using Grok, which is X's AI offering?
No, they're not really part of the big conversation, but there's a lot of money going into them and investing in the theory that Tesla and X and Elon Musk's various companies will deliver on AI in a way that other companies might not.
And that could be a very bad bet.
Or it could be that open AI is a bad bet.
You just don't know.
Aaron Ross Powell, I mean,
there's a lot of these companies out there that are making serious money.
You know, this is not entirely kind of a jam tomorrow thing.
It's not kind of, you know, buy this stock and just hope, you know, buy it basically as a bet and a gamble and see if it pays off.
These are real companies doing
real things.
I just think some of it is a bit excessive and
the extent to which it really does drag along the rest of the US stock market is for me a point of vulnerability rather than a point of strength.
But
I don't think this is necessarily going to crash like the
late 90s or early 2000s thing.
But I do think it's sensible to just try and stay slightly grounded along the lines of what is this technology actually doing.
Aiden, I don't think we're going to put this debate to bed.
We are never going to solve this debate, but so help us, we have tried.
Speaking of debates, we're going to be back in just a second with Longshore.
Speaking of alternatives, PGIM's monthly podcast discussing trends and strategies in alternative investing.
The history and evolution of the Australian superannuation funds.
It's interesting in so many ways.
They actually mandate savings into retirement plans, and they've become some of the biggest global investors in private markets now.
And their members have seen tremendous growth of their portfolios.
Hear the full conversation on PGIM's podcast, Speaking of Alternatives.
Okie dokie, it's time for long short, that part of the show where we go long, a thing we love, or short, a thing we hate.
I'm in charge, so I'm going to go first.
I am long, a comment from Mark Austin, who's a very prominent lawyer in the city of London and a very much a kind of city of London big Whig.
And he was talking about the Mansion House speech from the UK Chancellor yesterday.
She gave her big kind of annual speech to the great and the good of the city of London.
And some people were hoping that this would result in a massively impactful set of reforms for the city that could really like jazz up the stock exchange or whatnot.
And some people were a little bit disappointed by that.
But Mark Austin said, basically, you may be a little bit disappointed by some elements of this reform agenda, but, and I quote, we need to keep on shaking the snow globe.
That is my favorite quote.
It's a good one.
So shake your snow globe, Aiden.
What is your long or short?
I'm long stone fruit.
Much less serious topic.
It's almost stone fruit season.
You know, the cherries are coming out, the plums, and I'm just enjoying eating some delicious cherries before I go to bed every night.
It's been great.
Can I just strongly advise you not to overdo it?
Because this can go quite wrong if you eat too much stone fruit in one go.
So I don't wish to sound mumsy.
I'm just saying please don't over consume stone fruit.
You'll regret it.
Right, listeners, we're going to be back in your ears next week, so listen up then.
Unhedged is produced by Jake Harper and edited by Brian Erstadt.
Our executive producer is Jacob Goldstein.
We had additional help from TOFA 4hes.
Cheryl Brumley is the FT's global head of audio.
Special thanks to Laura Clark, Alistair Mackey, Greta Cohn, and Natalie Sadler.
FT Premium subscribers can get the Unhedged newsletter for free.
A 30-day free trial is available to everyone else.
Just go to to ft.com/slash unhedged offer.
I'm Katie Martin.
Thanks for listening.
The FT Weekend Festival returns on September the 6th at Kenwood House Gardens in London.
Register today for debates, tastings, QAs, and more.
Speakers include Stephen Fry, David Bediel, Nikolai Tangan, Nick Clegg, Tim Harford, and many, many more, plus your favourite FT writers and editors.
Register now and claim 10% off using the code FT Podcasts.