Baron Discovery Fund | Q3 2026

In the third quarter of 2026, Baron Discovery Fund (the Fund) declined 4.16% (Institutional Shares), outperforming the Russell 2000 Growth Index (the Benchmark) by 5.23%. This reversed the trend of the first half of 2026. Investors sold the momentum “AI winner” stocks and rotated into industries such as software and certain pockets of Health Care. The Fund is underweight “AI winner” stocks and the Momentum factor, and overweight software, so the reversal helped it recover some of its year-to-date underperformance.

The Fund still trails the Benchmark this year despite solid overall stock selection. At the end of the third quarter, the Fund was up 1.96% year to date, 8.75% behind the Benchmark. As you can see in the chart below, stock selection contributed a positive 2.0% to year-to-date relative performance but that was more than offset by the Fund’s underexposure to the Momentum factor (-3.6% impact) and underexposure to “AI winners” and overexposure to software and other “AI losers” (combined -4.6% impact).

The Benchmark’s year-to-date return includes a 3.3% contribution from Bloom Energy and Credo Technology. Both were in the Benchmark on January 1, 2026, but with market caps above $24 billion, they were too large for the Fund to have considered at that time. Both were removed from the Benchmark on June 26, 2026 when Russell concluded its annual reconstitution. Subtracting the contributions of Bloom and Credo from the Benchmark’s return, the Fund would have underperformed by 5.4% for the first nine months.
 

The “AI winner” trade falters

Our second quarter letter warned about the narrow and crowded “AI winner” trade of the first half of the year. We believe that AI is a revolutionary technology, but our view was that these stocks had risen too far, too fast. We equated the irrationality that we were seeing in “AI winner” stocks to the 1990’s dot-com bubble.

We explained that the AI buildout would proceed in fits and starts, following an S-curve rather than a straight line. “AI winner” stocks, however, were priced as if the growth would go “straight up and to the right” for an extended period.

The publication of our letter coincided almost exactly with the peak in AI exuberance. Starting in July, the air came out of the “AI winner” trade, and various semiconductor and AI “picks and shovels” stocks fell sharply. This pullback happened despite the large hyperscalers raising their AI spending forecasts for both 2026 and 2027.

Not every software company is an “AI loser”

As the “AI winner” trade stalled, our software holdings finally began to recover. Our entire first quarter letter explained why these companies are AI winners, not AI losers, despite market fears that AI would eventually disintermediate all software. In the third quarter, investors began to recognize what we have long believed: AI can help some software companies widen their competitive advantages and accelerate revenue and profit growth. Our software holdings rebounded strongly from extremely oversold levels.

We believe software stocks have significantly more upside. Despite the “SaaS-pocalypse” earlier this year, our companies kept executing their business plans. Their second quarter earnings largely met or exceeded our internal revenue and profit forecasts, contradicting the fear that AI would hurt software growth. The stocks remain undervalued in our view, and despite the recent recovery, most are still down year to date. We think the recovery in software has just begun and has more room to run.

Strategy and Positioning

Over the last several quarters, we have been vocal that chasing the momentum of the “AI winner” trade would be the wrong strategy, even though not doing so hurt our performance. This quarter’s sharp reversal in the trade is exactly why we felt so strongly. We believe a consistent, methodical process can produce outperformance in up markets and preserve capital in down markets.

The final chapter of the AI infrastructure build-out trade has not yet been written. Many of the stocks that have benefitted from this theme are cyclical rather than secular growth businesses, and we try to take a long-term approach by investing in secular growth companies. We are excited by the upside that AI can bring, but we are intensely focused on the risks as well.

Specifically, we see two key risks. First, capacity could expand too far on optimistic demand assumptions, and second, AI subscription revenue could fail to deliver an adequate return on the capital being spent. We are also seeing early signs of stress in the financing markets. Equity investors are demanding greater margins of safety, as shown by AI-related company IPOs being pulled for lack of demand, and credit investors are demanding higher spreads and tighter covenants (debt financing is a key component of the AI data center build-out complex). We will continue to rely on our process to navigate these risks.

Just as importantly, because the Fund looks to all sectors of the economy to earn its returns, the reversal of the software “pain trade” is not the only source of recent and future returns. Our Health Care holdings made the most significant contribution to the portfolio in the quarter, adding 4.33%. The Health Care companies that performed were each company-specific situations, in which improving fundamentals were catalyzed into higher stock prices. Identifying these long-term variant perceptions, where our view differs from the market’s, is the core of our approach. We discuss two of these stocks below. We are excited about the prospects for our Health Care investments and expect to share more in our fourth-quarter letter.

Looking forward, we are also very optimistic that our aerospace & defense (A&D) investments will bear fruit as they are both misunderstood segments and are trading at extremely depressed valuations. As it relates to the recent underperformance of our A&D holdings more recently, please see our longer explanation on why we think the market is wrong and we remain constructive on the sub-industry in this quarter’s Baron SMID Cap ETF shareholder letter.

Shares of CareDx, Inc. contributed meaningfully to third quarter performance. CareDx sells diagnostic testing services which detect early rejection of heart, kidney, and lung transplants. As we noted in our letter from last quarter, the company was waiting to hear from the Centers for Medicare and Medicaid Services (CMS) MolDX Program on a draft local coverage decision that, if finalized, would have significantly limited reimbursement for CareDx's tests. Our position has been that the outcome would not be as dire as what was being priced in by the market. We started to see some share price benefit from sentiment improvement surrounding the CMS decision in the second quarter, and in the third quarter we received a best case scenario result for reimbursement of kidney, heart, and lung transplant tests. This led to a nearly 130% return in CareDx shares in the quarter and over 2.5% of performance contribution to the Fund. This outcome is a prime example of how our fundamental research can lead to exceptional returns based on variant perception versus the market.

Tempus AI, Inc. is an intelligent diagnostics and health care data company with two synergistic business units: Genomics Testing and Data Solutions. The data generated from Tempus’ diagnostic tests enhances its data business, as Tempus combines its genomics data with clinical patient data from its oncology partners, to generate an enormous proprietary multimodal dataset. Shares were up meaningfully in the quarter on earnings and full-year guidance that beat consensus. Management was very bullish about the data business, noting a very strong biopharmaceutical pipeline. It also acquired Personalis, Inc., a company that specializes in minimal residual disease (MRD) testing to track cancer recurrence. MRD testing is a large market and fits nicely into the Tempus testing and data portfolio in our estimation.

Shares of Dynatrace, Inc., a provider of AI-powered observability software, contributed to performance as investors increasingly recognized observability as a primary beneficiary of enterprise AI adoption. Agentic applications expand the surface area that must be monitored: more services, more dependencies, more failure and security points, and far more telemetry per transaction. Because Dynatrace prices on consumption rather than seats, that complexity converts directly into revenue. Management disclosed that customers running AI workloads consume at 1.5 times the rate of those that do not, and the number of customers observing AI and agentic workloads grew sharply. Fundamentals confirmed the theme: annual recurring revenue (ARR) of $2.14 billion accelerated to 17% constant currency growth, net new annual recurring revenue rose 66%, and guidance moved higher. We retain high conviction, viewing observability as a durable picks-and-shovels budget line in the AI build-out.

Another company we’d like to highlight, though not among the top five contributors, is Rubrik, Inc., a cybersecurity and data resilience platform. The company’s shares contributed to performance as a rising tide in cybersecurity spending lifted the name. We added Rubrik to the Fund in the second quarter at the depths of the "SasS-pocolypse," when we believed the market had gotten it wrong on this company. A string of high-profile agentic AI security incidents this year has pushed cyber resilience and data recovery to the top of CIO and CISO priority lists. Rubrik’s platform unifies backup, identity, and recovery to enable rapid recovery from breaches, making it a direct beneficiary of that budget shift. The company’s fiscal second-quarter results validated this theme: net new subscription ARR reaccelerated to 35% year-over-year growth, roughly 27% above consensus, while subscription ARR contribution margin expanded to 14% from 9.4% a year ago. Management raised full-year guidance across every metric by more than the size of the quarterly beat. We maintain high conviction in Rubrik’s long-term prospects as enterprises continue to prioritize resilience alongside prevention.

Forgent Power Solutions, Inc. is a leading manufacturer of electrical distribution equipment for data centers, the power grid, and industrial applications. The company specializes in low- and medium-voltage equipment and focuses on custom, “engineered-to-order” products, which make up more than 90% of revenue. Larger competitors generally concentrate on higher-voltage and standard products. The stock declined during the quarter as sentiment turned against the broader AI infrastructure group, even though Forgent reported results and guidance well ahead of market expectations. Bookings reached $1.5 billion in the quarter, far above its current revenue run-rate.

Forgent is taking share by offering customized products with industry-leading lead times, and it is only beginning to tap its opportunity, having just signed its first master service agreement with a hyperscale customer. The company has invested heavily in capacity and people to support future growth. Given strong demand, its planned manufacturing footprint has grown to $5.8 billion, up from prior levels. We believe many years of strong growth and margin expansion as Forgent grows into that footprint.

Loar Holdings Inc. is a specialized aerospace components manufacturer focused on highly engineered, mission-critical products. Shares declined amid growing negative sentiment toward the aerospace aftermarket as elevated fuel prices related to the Iran war weighed on the outlook. We expect Loar to outperform peers, supported by latent pricing power and a strong new product pipeline. The company has not fully utilized its pricing power in stronger market conditions, providing flexibility in a more challenging environment. We also expect new product growth to accelerate as customers increasingly seek out leading suppliers to solve complex problems.

Shares of Mercury Systems, Inc. declined amid increasingly negative sentiment toward the defense sector, driven by concerns about the implications of a potential change in congressional control following the mid-term elections. Mercury supplies key, flagship national defense programs, including missile defense systems such as Patriot, electronic warfare systems, and ruggedized, high-speed data processing systems. We believe the company is positioned for accelerated growth for the foreseeable future as the programs it supports ramp production considerably in the next few years.

As it relates to the recent underperformance of our A&D holdings, please see our longer explanation on why we think the market is wrong and we remain constructive on the A&D sub-industry in this quarter’s Baron SMID Cap ETF shareholder letter.

The top 10 stocks represented 29.0% of net assets, in line with historical levels.

During the quarter, we initiated a position in Jersey Mike’s Subs Inc., a franchisor of fast casual, submarine-style sandwich restaurants. It is one of the largest and fastest-growing limited-service restaurant brands in the U.S., with nearly 3,400 stores across all 50 states and Canada, nearly all franchised. The system generated $4.3 billion in sales in 2025, and its loyalty program has more than 12 million members.

Jersey Mike’s pairs a premium, made-to-order product with one of the simplest and most efficient operating models in quick service. Subs are made in view of guests on a standardized 14-foot line, supported by a distribution model for core ingredients, and a hot grill menu (including Philly cheesesteaks) differentiates it from other sub chains. The brand ranked number one in quick-service customer satisfaction in the American Customer Satisfaction Index for 2026 and has 90% aided brand awareness, both rare for a chain of its size. The store format reinforces the advantage: new units are small (1,200 to 1,400 square feet), need no drive-thru, and cost about $515,000 to build. With average unit volumes of $1.4 million, franchisees earned a 2.6 times sales-to-investment ratio and roughly 42% cash-on-cash returns in fiscal 2025. A strong owner-operator culture, including about 360 hours of franchisee training and existing owners accounting for more than 90% of the development pipeline, together with a $200 million national ad fund, would be very difficult for a competitor to replicate at scale.

We estimate systemwide sales to keep growing at a double-digit pace over the next several years, following an 18% annual growth rate from 2021 to 2025. Growth should come from 8% to 10% annual unit growth toward a long-term potential of 7,500 U.S. and 15,000 global stores, positive same-store sales (20 consecutive years, 3.2% in 2025), and average unit volumes that have compounded at roughly 6% a year since 2006. As a capital-light franchisor, the business generates adjusted EBITDA margins above 50%, with capital expenditures of just $4 million in the first half of 2026. We believe the shares are very attractively priced and offer the opportunity for strong appreciation from here.

We also purchased Reformation Inc., which went public earlier this year. Reformation is a digitally native womenswear brand selling dresses, tops, bottoms, sweaters, denim, and accessories at prices from about $40 to $1,500. It generated $507 million in net revenue in 2025 and had about 1.14 million active customers. About 90% of revenue comes from its own website and 70 stores, with the remaining 10% from wholesale partners across 142 doors.

Reformation’s edge is how it develops product. Many apparel brands work with lead times of 12 months or longer, forcing them to guess at trends and rely on promotions. Reformation designs in-house and produces through a diversified supply chain, including a Los Angeles factory and partners across North America, delivering about half of its products within 60 days of ordering. New styles are tested in small quantities twice a week online and weekly in stores, and the team reorders winners within hours. As a result, about 90% of direct-to-consumer (DTC) apparel revenue in 2025 came from styles informed by customer data, and roughly 80% of DTC revenue from 2021 to 2025 came from full price sales. This constant newness has built a loyal following: 77% of customers cite Reformation as a favorite brand, about 75% of new DTC customers in 2025 were acquired outside paid digital channels, and nearly 70% of DTC revenue came from returning customers.

Based on our estimates, we see mid-teens revenue growth over the next several years, building on a 19% annual growth rate since 2023. New store openings are a key driver. Full-price stores open for all of 2025 generated $2.7 million in revenue each, and new stores typically pay back their build cost in under two years. Stores also lift the broader business. In markets with a store, active customers grew 16% annually versus 11% elsewhere, and DTC revenue grew 20% versus 14%. Customers who shop both online and in stores generate 3.1 times the revenue of single channel shoppers. Reformation is present in only half of the top 50 U.S. metro areas and sees a path to more than doubling its store count over five years. Further growth should come from new categories (dresses have fallen from about 60% of DTC revenue to roughly one third) and international markets (18% of revenue). We believe Reformation has a differentiated model with a long runway for growth, and that its shares are attractively valued at a discount to apparel and e-commerce peers.

We participated in the IPO of Lyntris Inc., a fast-growing proprietary defense solutions provider with a 90% sole-sourced portfolio exposed to many of the nation’s key defense offerings. Built via the acquisition of a few, smaller mission critical suppliers, Lyntris has become one of the best-in-class merchant suppliers that will benefit from expected rapid increases in production of important programs across the maritime, air and missile defense, and space submarkets. The company is involved in over 200 programs serving the large defense primes as well as newer defense tech players. It delivers complicated, difficult to create sensing hardware (RF and antenna systems), AI-driven data fusion (that combines multiple data sources into a single picture), and mission systems (including AI-enabled command and control systems) that are engineered to perform in the most demanding environments and are built on open standards using modular architecture. Lyntris is working on programs such as missile defense radar systems (LTAMDS) and communications systems meant to enable navigation on and direct communication from the moon (NASA LEGS program). Management is highly capable, having previously grown the two core companies that were put together to form Lyntris. We believe Lyntris can drive revenue growth of around 30% for the next several years alongside expanding EBITDA margins into the low 20% range with above average and improving free cash flow conversion. Cash flow will grow over 40% next year, yet the company trades at only 11 times next year’s EBITDA. It trades at a low multiple because it is unknown and unloved in the current environment where investors are waiting to see how mid-term elections affect the defense budget. We think that these concerns are overblown and that Lyntris is a compelling multi-bagger investment at these levels. We look forward to watching its story unfold.

We sold our position in Wingstop Inc. during the quarter to allocate the proceeds into other restaurant companies where we felt there was better upside potential.

We trimmed positions in Rubrik, Inc. and CareDx, Inc. after significant share price gains (up 43.4% and 129.7%, respectively) lifted them to outsized weightings in the Fund. They remain among our larger positions, and we continue to see attractive return potential in both.

2026 has remained a challenging year despite the recovery in performance during the third quarter. That recovery has continued into the fourth quarter as the unwinding of the “AI winners” trade has extended, with the Fund continuing to outperform the Benchmark quarter to date through October 16, 2026. The most recent catalyst was CoreWeave, where concerns about the financing behind the AI infrastructure buildout drove a sharp selloff in semiconductors and other AI winners. We are competitive people by nature and hate to underperform. That said, we believe our process produces consistent outperformance over a full market cycle.

Our approach is to think long term and focus on sustainable competitive advantages and strong management teams. We believe that formula will stand the test of time. It can lead to periods of underperformance when our style is out of favor, but we think focusing on company fundamentals with a risk management overlay is the right way to preserve capital in market downturns. In the meantime, we will continue to use this volatile environment to find new investments that we believe can deliver outsized returns in the future.

Thank you for your trust in us!