Michael Baron arrives at the conversation1 with Matt Zeigler of Excess Returns sounding less like a portfolio manager defending a thesis than one eager to share it. "There's so much going on," he says, and the co-president of Baron Capital, steward of $69 billion in client assets, means it as good news. The excitement is not a mood. It is a reading of a market that, in his view, is mispricing both the obvious winners and the supposed losers of artificial intelligence.
Growth, Everywhere but Where Everyone Looks
The first surprise is structural. Baron Capital is a pure growth equity shop, yet the firm tends to run underweight technology. The reason is correlation. "If you're just focused on AI and just focused on technology, they tend to trade together. They tend to go up together and tend to go down together. And it's much too volatile for the average investor to stomach." The team instead hunts for growth in financials, real estate and consumer businesses that most growth managers ignore.
The firm does not chase short-term scorekeeping. "We don't say we're going to try to beat in any one given quarter or in one given year." Over the long haul, Baron reports that roughly 97 to 98 percent of its strategies have beaten their benchmarks since inception, with 92 percent of funds in the top quartile of their categories.
Zeigler draws out the point. "I love that you brought up the point about the breadth of growth opportunities, because I think that gets lost in the conversation."
Not a Bubble, Not a Free Pass
Asked whether AI is a bubble, Baron is blunt. "It clearly is not." To be clear, real does not mean universal. Some companies will fade; others will excel as enablers or users of the technology.
The market, he argues, has lost its ability to distinguish between them. "Right now people are confused, concerned. They're painting everything with the same brush." Software, professional services and data businesses have been lumped into the loser column, and their valuations show it. "You know, they're on sale." Shopify, Guidewire, Gartner, FactSet and MSCI have seen multiples compressed even as growth runs from high single digits into the 20 percent range.
The defence is proprietary data. "AI doesn't work in a vacuum." It "only runs with data," and companies sitting on decades of unique customer and industry data are, in Baron's words, "enabling companies to use AI, not being disrupted by it."
The Railroad to Space
The Musk holdings illustrate the firm's long-horizon conviction. Baron Capital spent about three years doing diligence before building its Tesla position between 2014 and 2016, when the company produced roughly 30,000 vehicles. The mission, Baron argues, "has nothing to do with transportation. It has to do with energy and energy abundance and sustainability of things." Autonomy is the next leg, and in his view Tesla "is the only one who's really bringing the physical and the digital world together."
On valuation, he notes that "most people are thinking about valuation because they're thinking about today." The firm's average holding period is "8 or 9 years," and on Tesla, "we're at year 12 and we have no signs of selling this."
SpaceX elicits even more enthusiasm. "I mean, SpaceX, you get so excited you can't sleep sometimes." Launch costs per kilo have fallen from NASA's roughly $50,000 to about $1,000, with Starship aiming for the hundreds. "You now have the railroad to space," and roughly 90 percent of mass to orbit already rides on Falcon. "So this is a company that competes with countries, competes with China." Setbacks are part of the process. Echoing Musk, "there's nothing in physics that prohibits what they're trying to achieve, and therefore eventually they'll achieve it." The ambition extends to orbital data centres: "You're going to do inference in space and you're going to do training here on Earth."
On key-man risk, he concedes, "We're definitely investing in Elon," but stresses the bench. "There's a next man mentality over there."
Time, Balance and the Hardest Trade
Zeigler frames time as an edge. Baron agrees. "Everyone wants things now, right?" Investors punish capex cycles, which is exactly when the team buys.
Balance matters because leadership rotates. AI winners outran AI losers by 28 points in 2024, 15.5 in 2025 and 47.5 in the first half of 2026. Then, over the past two months, "it reversed." The lesson is simple. "You need balance in portfolios."
Position sizing gets the same candour. "It's complicated. It's hard." Tesla grew from 8 or 9 percent of Baron Partners Fund to about 40 percent. Critics called it reckless; Baron counters, "It was risky when we made that first investment." Rather than sell to an arbitrary weight, the team reduced leverage and added stable growth names in financials and real estate. Meanwhile, the Mag 7 are, in his words, "highly correlated and that's a big portion of the portfolio" for index investors.
Selling is reserved for deteriorating competitive advantage, never timing. "The second hardest thing to do is sell, and the hardest thing to do is to buy back after you sold." His father's rule applies: "don't get cute. Don't wait for it to bounce."
People Over Moats
Baron respectfully departs from Buffett's preference for businesses any idiot could run. "That's not us." Judging management is intuitive. "You knew who the smart kid was." He recalls pulling off the Long Island Expressway at the start of COVID to take a call from Hyatt's Tom Pritzker, whose balance sheet discipline meant "he was built for something like this." In turbulence, the prescription is primary research. "You need to get on planes."
Judgment Is Not a Commodity
The firm is becoming wrapper agnostic, adding ETFs, SMAs and CITs. "The mutual fund's a fabulous wrapper," but "we're not beholden to it."
His contrarian view: "I view AI going to make active management more important, not less." Because "AI is going to democratize Data, but it's not going to democratize judgment." Zeigler agrees: "And there's this myth that it will replace judgment."
The closing advice ties it together. "Curiosity leads to primary research, leads to conviction, which leads to these outsized returns."
5 Key Takeaways for Advisors and Investors
- Diversify within growth. AI and tech exposures move together; pairing them with growth in financials, real estate and consumer names reduces volatility without abandoning growth.
- Separate AI victims from AI enablers. Businesses with proprietary data, such as FactSet, MSCI, Guidewire and Shopify, may be mispriced as losers while still growing.
- Value the future, not the present. Long holding periods let compounding and optionality do their work; today's multiple rarely captures a mission-driven business.
- Manage concentration at the portfolio level. Trimming leverage and adding uncorrelated holdings can contain risk without selling a high-conviction winner to an arbitrary weight.
- Sell on thesis, not price. Exit when competitive advantage erodes, act decisively, and remember AI democratizes data but not judgment.
Footnote:
1 Returns, Excess. "They Beat All US Stock Funds Since 2003 | Michael Baron on the AI Winners Investors Miss." YouTube, 1 Oct. 2026, www.youtube.com/watch?v=KVCgmorfy50.
