Ausbiz: Navigating the AI Investment Boom: Risks and Opportunities

Well, let's get to what we're seeing now on Wall Street and AI and speculation about hype drawing investors into the sector. Sam Ruiz is joining us from T. Rowe Price. Sam, good to catch up with you again.

Good to see, Andrew.

Well, we, I mean, we've spoken about this a number of times. If we talk about an AI frenzy, if you like, I mean, is that justified because as we know the talk of a bubble, what are your thoughts?

Yeah. So I think we've been pretty positive on AI for over a year now. The last quarter though, I would say, really put us into a new trajectory. I guess I'd say there's two things.

It was the retail investor appetite for some of these baskets and it even extended beyond AI.

So seeing really big almost parabolic moves in some areas such as even space stocks even got crypto baskets doing incredibly well.

Quantum computing's another one fun fact for your quantum computing stocks, these are companies that really we don't even know for the next 5 plus years if they will be successful.

But those stocks are up about 450% since Trump was elected.

So things are moving very, very quickly.

On top of that, we've had this deal frenzy really kick into market.

So the last quarter we had open AI basically getting out there with multi $100 billion deals, some of these even vendor financed.

And it wasn't just the extent of these deals, but it was how they were structured, which really creates now this circular finance intertwined system in terms of the investment of AI, which now just gets us into a phase where we think it's important to be bubble watching.

Not necessarily saying we think that we're at the point where the bubble is going to pop, but certainly it feels to us like we've now entered some element of a bubble.

Yeah, it's a question of how far we go, but also how you navigate a bubble essentially, isn't it 100%?

So if we were to take back I guess the late 90s into the 2000 pop, we think if we were to give an analogy, we're closer to around mid 1998.

And that's because there is still a little bit more duration we think to play out in terms of this investment cycle.

Some of the commitments from OpenAI and others are really planning out over the next 2-3 even up to 5 years.

So some of that money hasn't yet been spent, but we start to look at the capital intensity of some of the even hyper scalers which are really well funded eating into the cash flow.

Question marks about the return on investment is something we're certainly worried about, but it's really a big question mark around at what point do people figure out maybe we've potentially built too much AI capacity or the returns aren't there or something underneath that something triggers a pop.

Now one of the things we're really navigating and I guess probably about 3 or 4 things we're actually watching is does the Fed effectively change course?

Now what would really change sentiment and the direction in the late 90s was the Fed eventually hiking rates.

We're obviously not there yet, but if they say that there's a change of sentiment that would be bad.

If inflation data surprises to the upside, that would be bad.

And obviously that's connected to rates.

We've got to watch employment and that's something that would really start to spook people because there is a level of top line economic activity that needs to fulfil expectations for the revenue and cash flow that supports this investment cycle.

And on top of that, it's really the narrative around AI in general.

So if we started to get data points such as there's not enough electricity to fund or to actually fuel and power some of these data centers with where expectations are, any type of sort of reaching of investor optimism we think could be quite negative.

What I've also, I mean, does the trade between the US and China also play a part here?

Because in fact we've just heard that the Trump administration now considering are putting curbs on software exports to China.

Yeah, I mean that is something that is I would say more for broader market sentiment.

You can see the market really gyrate around, you know, we're friends with China now we're going to put more sanctions on.

I think there's a little bit of ebb and flow some sort of pre meet deals going on right now or almost just more the jawboning and the sentiment heading into a deal.

So that's something I don't wouldn't say we have any insight into how that gets resolved.

But if that significantly only pivoted in a in a worse direction, I think you would see the stocks which have gone to the most elevated valuations and have the highest expectations really get hit the hardest.

One thing I would, I would say on the back of that, it is really unusual when you look at equity market drivers, you can look at factors that drive the market, you can look at quality, momentum growth, etcetera.

We've actually seen risk and low, sorry, high risk and low quality stocks significantly outperform over the past 12 months now and they're in their 90th percentile since 1992.

What that effectively means is if you were to invest over that period of more than sort of 30-35 years, only one in 10 times have these high risk, low quality stocks actually been the ones to bet on.

And following that period of outperformance is generally been quite bad for them.

So this is just an indication of how far the markets getting the hype, the sentiment, the speculation.

But to your point on Trump and she effectively US and China, that's the part of the market we'd be worried about more if optimism was effectively eroded from something like that.

What are you avoiding then and talk about that that craves the hype there.

Which sort of stocks are you avoiding?

Yeah, I'd say there's a, there's a basket of meme stocks in general that we're avoiding.

Two high profile ones that we're on record saying that we're really leaning away from now would be Palantir and Tesla.

Now these are both profitable companies, but when we talk about high valuations mean high expectations and high risk of disappointment, both of these stocks trading around 200 times earnings now.

Now, just as a result, overnight wasn't actually too bad, but it did disappoint some of the higher and more lofty expectations.

And when you have valuations now, I'll just step back and say both these companies have a very good, whether it's a consumable good or service that consumers will want in the future.

But at that level of expectation and valuation, they're the ones where we just think we want to shy away from be with more reasonably valued companies that we think have durable moats and that cash flow sort of visibility into the future.

All right, what are those companies, the ones durable with moats?

Yeah, I mean there's, there's a number of them.

They feel like there's few of and there's few of those in the market today in the AI segment specifically, which is an area I'd say we're more cautious on.

And we're kind of in the middle now, bubble watching, but aware that this deal frenzy means that there could be a lot more to be spent in the future.

We're leaning away from companies that are effectively spending the money.

Now we've talked a lot in the past about the hyper scalers that are spending a lot of their very strong balance sheets in cash flow and AI, but they effectively now seem like they're in an arms race and might end up commoditizing each other, basically building as much as they can out of competitive imperative, but that might erode their own margins.

So we want to lean now more to the companies where that money is being spent on, but where there's much less risk that someone can come in and do that same thing that they do.

So there's four specific companies, all very well known that we lean more favorable on in AI right now.

That is TSMC, NVIDIA.

It's also Broadcom and AMD.

Now if you look at those four stocks, three of those are effectively chip makers.

We all know that NVIDIA is the one that has effectively the best GPU in the world, but AMD is the second best and they're trying incredibly hard to have a competing chip.

Then you have Broadcom, which effectively makes custom chips.

They're not as good for broad use across all AI uses, but if you get to 1 specific use case, they can be incredibly productive and very, very strong.

Now the 4th one there is TSMC, which is effectively leading edge monopoly of making these chips.

And if you need to make them, the only person you can really use is TSMC.

So all four of those we think are in the box seat still as AI momentum stays strong, it's a safer way to play it.

And one data point we have today is that the ecosystem of money being spent on AI chips is currently around $45 billion USD.

We think that that's accelerating maybe in the next two to three years, three years as high as $500 billion and this is dreaming big, but sort of by the end of the next 10 years, it could be as large as $1 trillion.

So how you play that is in the companies that you think have something someone else can't really do and are in the box seat to monetize that.