The Midwit Graph: Five Lessons from Dom Rizzo

Jack opened our interview Dom Rizzo with an admission about his prep. He wrote his questions during a big correction in AI stocks. By the time we got to the interview, they were rallying again. Such is the nature of the market today. And as Joe Weisenthal and Tracy Alloway say on Odd Lots, we had the perfect guest for that type of environment.

Dom runs T. Rowe Price’s Global Technology Equity Strategy and the firm’s technology ETF, and he joined Jack and guest co-host Kai Wu of Sparkline Capital. He took over the global technology strategy on December 1st, 2022, one day after ChatGPT was released. “Sometimes you get lucky on timing.” He went AI on and semiconductors on, and the rest of the conversation was about navigating a period where that luck has turned into potentially the biggest technology revolution of our lives.

Near the end of the hour, Jack asked our standard closing question, the one lesson Dom would teach the average investor. Dom’s answer reminded us of a trap we all often fall into. The midwit graph is the one shaped like a bell curve, with the so-called dumb guy at one end, the really smart guy at the other, and the average person who thinks they’re smart on the hump in the middle. The two ends think the same thing. The person in the middle overthinks it. Dom’s version of what both ends already know was simple.

Buy accelerating fundamentals.

But the implementation is less important than the lesson itself. Often doing the right thing in investing is simple. When it became clear that AI would be world changing, the best analysts worked out through detailed analysis that we would need a lot more semiconductors. But a simple thought process would have led to the same conclusion. Those who overanalyzed the situation in the middle were the ones who missed the simple reality.

Jack said that gives hope to dumb guys like him, and Dom backed him up with Buffett. “You don’t need the highest IQ to be in this business. You need a high enough IQ.” Dom runs one of the biggest technology strategies in the country, in what he’d just called maybe one of the greatest reflexive cycles of all time, and his one lesson for the average investor was the beginner’s answer, given on purpose.

Dom has a speech for the bubble question, and he delivered it on request.

AI has the potential to be the biggest productivity enhancer since electricity. Productivity-enhancing technologies come with speculative bubbles. My job is not to miss bubbles, but to navigate them responsibly for our clients by trying to capture upside and blunt downside.

“I know that’s a mouthful, but I’ve practiced it a few times.” What changed this year is the funding. Last year his job was easy, because the build-out was “funded by the free cash flow of the most successful organizations of all time.” Now one of the most profitable companies on earth (Google) has issued $85 billion of equity to keep investing and we are seeing capital coming from a variety of sources that are not free cash flow. “We are in the capital cycle part of the build-out.” That’s why his main worry in July was reflexive, whether a price correction could get material enough to freeze the equity and debt markets that fund the spending. Do the math and the funding gaps, a few hundred billion per company, are manageable against market caps in the trillions. But belief carries the load. When Jack asked the show’s other standard closing question, about a view peers would dispute, Dom said “the power of reflexive cycles is underestimated” and called this possibly one of the greatest reflexive cycles of all time. Then he gave us a reading assignment. If you want one book on stock picking, it’s Soros, The Alchemy of Finance. Situational Awareness, when it blew up this summer, had the thesis right and the leverage wrong, and as Dom put it, leverage helps on the way up and hurts more on the way down.

Some have argued semiconductors are no longer cyclical. You might expect a tech fund manager to agree. But Dom doesn’t buy it.

Semiconductors are always cyclical and will always be cyclical. Let’s be very clear. We just happen to be in a great up cycle right now.

There will clearly be a correction someday. Semis sit at the end of the bullwhip, where Google sneezes and a small-cap optical stock catches “not just a cold, potentially almost bankruptcy.” What has changed is the driver. Software and the internet scaled massively with minimal capital, which was the beauty of that era. AI is not that. “AI is token manufacturing, and token manufacturing is capital intensive.” The scaling laws are the reason, since throwing 10X compute at a problem buys roughly 2X more intelligence, and there’s been no evidence the scaling laws are ending. Ben Horowitz put AI penetration at about 3 percent, a number that seemed right to Dom.

The obvious objection is that intelligence has its limits. You’ve probably talked to someone with a 130 IQ and someone with a 140 IQ, Dom said, “and they may have both been a little boring.” His answer is to stop thinking in IQ. “Don’t think about it as IQ, think about it as task completion,” and a million use cases open up. The gap between talking to intelligence and delegating to it is the whole game, and it’s a gap measured in tasks, not IQ points.

Jack asked about the balance between productivity and job loss.

I think it’s dishonest to say that AI won’t displace some jobs. I think AI will clearly displace some jobs, just like all technology innovation has always displaced jobs.

Bank tellers gave way to ATMs. Email and Word ended the typist. There was even a human job called computer at one point. Tech revolutions have eliminated jobs. AI will do the same.

But the working world’s response so far isn’t unemployed engineers. Every software engineer he knows is 20 to 40 percent more productive and working the hardest they’ve ever worked, because there’s suddenly so much worth building. On net he expects more jobs, including the blue-collar ones already showing up in electricians and plumbers for data center construction. Keynes predicted his grandchildren would work a couple of hours a day, and we don’t, because “humans are status creatures, and we love building.”

He made the case with his own family. If he could tell his great-grandfather stepping off the boat at Ellis Island that his great-grandson would have a personal trainer and a job sitting behind a computer screen, “it wouldn’t even resonate.” Dom won’t pretend to name the new jobs. Asked what they’ll be, “I have no clue,” which he admitted isn’t a great answer, except the new jobs have popped up after every technology shift in history. He also granted that none of this is comforting to the person whose job is the one displaced. He’s not in the age-of-abundance camp and he’s not a jobs doomer. He expects high churn, accelerating GDP, and a lot of people doing work that doesn’t have a name yet.

Jack, a quant, wanted to know how a fundamental investor builds a portfolio around a revolution since quant strategies typically aren’t a good fit. Dom reminded him that isn’t 100% true. He thinks quants have historically done portfolio construction better than traditional managers, so he borrows from his quant team on beta and momentum exposures.

In terms of how he builds a portfolio, he tries to make his fund “the easy button in tech,” and his framework fits in one simple sentence: linchpin technologies innovating in secular growth markets with improving fundamentals at reasonable valuations. A linchpin technology is one that’s mission-critical to its customers’ success, companies like ASML and TSMC. He worries more about factor exposures and sub-sector weights than individual names, and his standing conversation with his risk team is about unintended bets. He’s happy to take bets, since that’s how alpha gets generated. He just wants to know which ones he’s taking, because as with Rumsfeld’s unknown unknowns, the risks you don’t know you’re carrying are the ones that get you.

The framework has a warning built into it.

Often the market will dare you to buy stocks ‘cause it will have a few of the factors, but not all of them.

Memory chips were his example, with decelerating fundamentals and single-digit PEs. “That’s the market daring you to make a choice between valuation and fundamentals.” And some of his best ideas, he said, screen terribly on his quant screen. That’s where the edge lies.

Dom prefaced his 1998 comparison honestly. “I was five years old in 1998, so this is not personal stock picking experience.” What he does have is a way of reading cycles that doesn’t depend on having lived them. Corrections inside a great up cycle can look just like the corrections that end them, so price action alone can’t tell you which kind you’re in. Fundamentals, on the other hand, can help. When the selloff came, his semiconductor numbers looked nothing like 1998’s, and that’s why he read it as “a 1998 style correction that results in an even stronger follow on” rather than a top. His favorite chart overlays the Nasdaq of the Netscape era on the Nasdaq of the ChatGPT era and put the moment at “literally the halfway point.” It’s a forecast, and his own practiced speech from the beginning already allows for being wrong, since the job is capturing upside while blunting downside. The method, not the conclusion, was the real lesson of the hour. A simple idea, honestly held, is “usually a great place to hunt.”

Watch the full episode here: