Semiconductor giants are powering the AI revolution

You can also listen to this podcast on iono.fm here.

SIMON BROWN: I’m chatting now with Izak Odendaal, Old Mutual Wealth investment strategist. Izak, appreciate the early morning time. In a recent article you put out about a number of issues around sort of AI, the capex boom, the spend you make a great point that, if anything, the AI capex boom is actually getting narrower. It’s not broadening out. We are seeing more concentration in the indices as it continues afoot from the really, really giant hyperscalers and semiconductor shares.

IZAK ODENDAAL: Yes. It is really the semiconductor shares that have performed phenomenally well. And now you have a company like Micron – which I think most South African investors have never heard of. Suddenly it’s worth more than $1 trillion.

And so yes, we’ve seen this divergence this year or, let’s say, over the last couple of months, between the semiconductor side – let’s call it the hardware side – versus the software side.

And what you’re seeing is essentially, I think, the market saying that on the software side, there’s actually a risk that AI could disrupt some of these business models – whereas in the past, of course, hardware and software kind of moved together.

The best place you can see it actually is just in the performance of emerging markets. If you look at the absolute flying stock exchanges of Taiwan and South Korea – where of course, most of the chip manufacturing is done – it is just phenomenal.

Read:
Big tech is suddenly impacting niche market for US dividends
High volatility ruins the mood for investors
$600bn AI splurge could keep US growth humming

SIMON BROWN: Yes. South Korea in dollars more than doubled this year. And Micron a 10-bagger in 12 months. And, to your point, I had never heard of Micron until about two or three weeks ago when it broke the trillion-dollar market cap.

You make the point; this is not a 1990s-style bubble in the narrow sense. There is profit here. These companies are making a fortune. SK Hynix, Samsung, Micron in their chip business, really are printing cash. But it is cyclical, perhaps, is the important point.

IZAK ODENDAAL: Well look, yes. It’s definitely not a bubble or a valuation bubble because, as you say, these companies are very profitable. The bubble might be in the earnings of these companies.

It might be that two or three years from now the data centre build out slows, the demand for semiconductors either declines or just peters out.

And then the prices have to come down to earth.

The margins of some of these companies are phenomenal, and that obviously is going to invite competition. As you say, historically semiconductors have been an extremely cyclical industry precisely because these kinds of demand and supply cycles are not always in whack. And then you end up with these big price cycles.

Read:
What the gold and AI booms mean for investors
Mag 7’s growth outpacing the S&P 500

SIMON BROWN: Absolutely. And we’re also seeing – I hadn’t thought around this part of it – you make a really great point around supply chain and the concentration in that. You’ve mentioned Taiwan, South Korea. That’s where we’re getting our chips from.

Helium is coming out of Qatar. ASML is the only group making those photolithography machines that everyone needs. There is a fairly significant concentration. And of course, this ultimately potentially gets political. We’ve seen the US CHIPS Act; we’ve seen potential China export bans. We’ve seen US bans on Nvidia chips. We’ve seen Intel stakes.

This starts to get quite geopolitical at the same time.

Read: Taiwan market cap tops $4trn on AI boom, overtaking UK

IZAK ODENDAAL: Yes, it’s interesting if you think about it. In the last 70 years geopolitics has been driven by oil. I think it’s safe to say that over the next decade or so geopolitics could be driven by chips.

Indeed there’s a great book by Chris Miller called ‘Chip War’, which kind of gets into all these details and all these countries trying to secure their own supply chains because it is a massively concentrated industry.

And there are very obvious chokepoints in this whole supply chain.

SIMON BROWN: And then for the investors out there diversification is always the way to sort of manage this. But that diversification perhaps in some ways is becoming harder than it looks. Maybe not so much for South Africans.

But if you have to diversify away from the US, you’ve a nice EM fund. And of course that EM fund is now probably a quarter, a third, and a half South Korea.

IZAK ODENDAAL: Yes, I think  the kind of underlying risk here is that the AI theme is driving the US market, is driving the EM market, as you say.

There’s a big, big slice of it now spilling over into corporate bond markets. It’s driving a lot of economic activity in the US – the data centre buildout. That’s all good and well while it lasts and no one knows how long it lasts.

All of these things are still up in the air –whether all these business models turn out to be profitable. There are just too many questions around AI. But obviously the more concentrated the whole market ecosystem becomes, as you say, it is more difficult to diversify yourself into areas that are not directly exposed to AI.

Read:
Africa startups turn inward as US AI boom drains venture capital
Wall Street is searching for AI winners across emerging markets

SIMON BROWN: And, if anything, there’s concentration. This mad rush to AI actually gets more high risk. I’m talking now of the mega IPOs. We’ve got SpaceX next week. We’ve got Anthropic filed confidentially, SpaceX probably around $2 trillion. Anthropic around $1 trillion. OpenAI won’t be far behind. That’s going to be another trillion-plus.

We’re nowhere near the end days of this. This is still ratcheting up.

IZAK ODENDAAL: Yes. And those IPOs I think will just drive further excitement and potentially keep this growing.

If you look back historically, though, often the big blockbuster IPO has been near the top of the cycle. I’m not suggesting that that is the case now. I’m saying that’s been the case historically, because obviously that’s the best time to list.

The best time to invest is when investors are clamouring for your shares and they’re willing to pay huge amounts of money to get hold of your shares. So I think that is another thing to bear in mind. I’m not calling the top. I’m just saying this is kind of behaviour that you get closer to the top and the bottom of the cycle.

Read:
Nokia’s 140% rally turns AI comeback into valuation puzzle
Nvidia rival Cerebras seeks to raise $3.5bn in US IPO
China will open its market to AI chips from the US, Nvidia’s CEO says

And of course, the other problem for investors is that, whereas with something like Nvidia we’ve benefited as it has  grown massively with these IPOs, a lot of the growth has taken place in private hands.

And so we, as public investors  are maybe not going to get the full benefit of that.

SIMON BROWN: That’s actually a great point. Nvidia – you could have bought it when it was a $100 million market cap. It’s now, you know, $5.5 trillion. These mega IPOs you are buying at already $1 trillion to $2 trillion. Most of that profit has gone to the private markets.

We’ll leave it there. That’s Izak Odendaal, Old Mutual Wealth investment strategist. Appreciate the early morning time.

#Semiconductor #giants #powering #revolution