Lyle SteinA Comment -- General Comments From an ExpertA CommentaryCOMMENTFeb 07, 2025

Every fundamental factor he's looking at indicates that we still need more compute. At the beginning of last year, we thought we were going to spend $350B in capex for 2026. Now that number's looking like $1T, going up to $1.3T for next year.

We're still seeing a significant amount of demand, which is not being met by the current supply in the market. For now, we still have a couple of quarters of solid demand getting ahead of supply.

Where you have to pay attention is as soon as the margin profiles come back down. You can have periods of glut. At the end of the day, these are more commodity-based assets.

Call him crazy, but he thinks we can buy software stocks again. He came out saying that a lot of applications in some of the horizontal software companies were going to be disrupted in a massive way. It's a function of a lot of alternatives being available on the market.

But you still need a lot of the software infrastructure companies. The ones that enable the AI agents were much more immune to the selloff of the SaaSpocalypse. Any companies you buy have to have an AI angle and have to be reaccelerating revenue.

Safe to get back into software, but it has to be on the infrastructure side.

People looking at these massive capex numbers see the spend side, but want to see the revenue side. Seeing explosive revenue from Anthropic and OpenAI. He's starting to pay much more attention to is the return on investment among the hyperscalers.

GPUs have a longer life cycle than people are expecting (9 years vs. an estimated 6). So the payoff period can extend much longer.

His point is that this has been a one-decision type of investment over the last 10 years. The compound is something like 22% over that time, but that's not a realistic expectation for investors. We've ridden a tremendous wave of AI, cloud computing, and electrification of the grid. If you look at the fundamentals of a lot of the NASDAQ companies, especially the Mag 7, they deserve to trade at high valuations.

That said, a lot of companies have come out of nowhere with big increases in market cap. No question, demand is off the charts for AI equipment. If that cools, we could face a period where (though some companies are doing amazing) the NASDAQ and the S&P could do nothing for years. It's simply because of the way the market's structured, with so many companies tied to the AI trade.

This worries him a bit. He wants to make sure his clients have reasonable expectations going forward. For a diversified portfolio set up for your retirement, the equity part of your holdings is looking at 8-10% a year and not 22%.

It's a good economy. Seeing lower unemployment, strong corporate earnings. When people spend $$ on infrastructure and AI, it blends into the entire economy. We're seeing the stock market be very strong, which leads to good vibes, rich people investing, and rich people retiring.

We're in for some pain at some point. He just can't tell you when. We're in the fourth year of strong markets for most of North America, double-digit returns this year. Sometime, the good news will start to fade. 

You need to be selective with your investments. There are so many wonderful businesses that used to trade at 30-35x PE, and now trading at lower valuations even though their growth is just as good. The market's focusing on a lot of nit-picky stuff that's not relevant to long-term investors.

All have done extremely well. Trading at one of the highest valuations of the past number of years. So are US banks, as is the stock market. Banks benefit from the beta of strong stock markets, but the reverse is also true.

The better question is should I be taking some $$ off the table (and that depends on your rick tolerance)? How much of my portfolio does it now represent? Money managers have  to follow rules on position size, but individual investors don't -- for them, it all depends on comfort level.

Right now he's attracted to companies that, for some reason, the market hates. Whether it's META, UBER, V, or NFLX. Some are at 52-week highs, but haven't done a lot over the last 4-5 years. Instead, money's been rotating into the hot areas. 

He sees so many opportunities in dislocated, high-quality companies. Growing really fast, but valuation is the cheapest it's been in a long time given the opportunities ahead.

He's held the hyperscalers as core holdings since 2015 and he still sees upside. The past quarter validated that with acceleration in the cloud business by Amazon, Microsoft and Google. Margins increased. But there will be more competition for AI services and prices are reducing for best-in-class models. Meta's in the doghouse from regulatory issues and are spending a lot of money but their core advertising business is on fire, which may surpass Google Shopify is using AI to accelerate its core offerings. As for software, Microsoft's Co-Pilot keeps getting better, while ServiceNow will build AI functionality across all its platforms. End users will use software they already trust, but will use AI.

According to his metrics, expect some choppiness in the coming weeks, but not devastating. The market tends to be softer from late-August to late-October on average. Any volatility should be bought.

He sold at the peak, around $5,400. It has since had a downtrend, is consolidating, so now is a good time to buy. Gold is back on track. 

He predicted a return to 72 cents, but now predicts a fall to 70 cents, perhaps 68, because we are the weakest G7 economy with little improvement. Also, our taxes our high compares to peers. He's been buying USD during this rally.

Nvidia is the next big earnings report, next week Wednesday, then there's Jackson Hole. This week will see just a lot of noise. From Nvidia he wants to see how this "leverage on leverage" of circular financing works, which recalls the leverage that led to the 2008 mortgage debacle. It's great that the rally is broadening and earnings keep rising. We're late in the cycle and concerned over bubble characteristics in the market, though overall he's bullish. Given current valuations, the 10-year forecast on the rate of return on the S&P is negative--but the peak may be two years from now.

They're like T-class mutual funds where you get a component of your return every year. This is very tax efficient; the full distribution in the current year is not taxable. For those seeking tax efficiency now and need current income.

BMO. But how much credit risk will you take--high yield or investment grade? What's your time frame? Historically, credit spreads are very tight, so don't take credit risk now. Because rates are backed up, he doesn't mind taking duration risk. But will it make him a total return positive in the next few years? Not sure.