Baron Discovery Fund | Q4 2025

Dear Baron Discovery Fund Shareholder,

"The fear of machines turning evil is [a] red herring. The real worry isn’t malevolence, but competence. A super-intelligent AI is by definition very good at attaining its goals, whatever they may be, so we need to ensure that its goals are aligned with ours... If we cede our position as smartest on our planet, it’s possible that we might also cede control.”
- Life 3.0: Being Human in the Age of Artificial Intelligence, Max Tegmark, pp. 67-68.

“Today, many of us choose to live in a form of pseudo-reality governed by algorithmically enabled individual experiences. Much of what passes for authentic experience today is vicarious and virtual... [W]hat began as a slow bleed of reality on the edges has now become a culture-wide destabilizing force. Reality has competition, from both augmented and alternative forms... In these new worlds, we are Users, not individuals.”
- The Extinction of Experience: Being Human in a Disembodied World, Christine Rosen, pp. 2-4.

“You do not rise to the levels of your goals, you fall to the level of your systems.”
- Atomic Habits: An Easy & Proven Way to Build Good Habits & Break Bad Ones, James Clear, p. 27

“Throughout my career I’ve had my antennae up, looking for examples of people who use systems as opposed to goals. In most cases, as far as I can tell, the people who use systems do better... To put it bluntly, goals are for losers. That’s literally true most of the time.”
- How to Fail at Almost Everything and Still Win Big, Scott Adams, pp. 31-32

The above quotes seem to be strange juxtapositions, but actually mesh together pretty well, and are very apropos to the world in which we live right now. As a society, we are at the cusp of massive advances in the realm of AI. We hope that this will be to the benefit of humanity, but right now is the time in which humanity will determine the parameters (goals) around which AI algorithms are ultimately limited in their negative abilities. At the same time, we are at a crossroads in the manner in which we interact not only with the world, but with the machines that are increasingly blurring what is real and what is artificially created. This is an existential rubicon demarcating the very essence of which we define ourselves as human beings. Are we merely “Users” who are social vassals to our algorithmically driven devices (and the social media virtual realities that they convey)? Or do we have the power to deterministically (via systems) lead ourselves to brighter humanistic and societal outcomes?

Laird and I are huge optimists, and we believe that the companies in which we invest (and more importantly the people that run them) are being systemically built for positive outcomes that will make every generation more prosperous, healthy, and intelligent than ever before. Goals may indeed be for losers (let that be the AI machines). Being systems-based might be the key characteristic that distinguishes humanity from the relentless goal seeking of AI models.

Critically, the process we use to manage Baron Discovery Fund (the Fund) is also systemic and is keyed to strictly adhering to our core long-term growth investing principles. Our target of earning 15% compounded returns over time is merely the fruit of our “system.” We invest for the long term in competitively advantaged companies that are well managed and have the potential of large untapped market opportunities. We then continuously monitor these companies and their competitors via our due diligence process. We allocate our overall portfolio via our risk process which includes balancing index sectors, types of growth, position sizing, and valuation parameters. We also spend many hours with the teams that manage the companies in which we invest, to gauge their mettle, morality and ability. No algorithm can substitute for this “live” research.

In the fourth quarter of 2025, the the Fund returned 0.19% (Institutional Shares), trailing the Russell 2000 Growth Index (the Index) by 1.03%. For the full year 2025, the Fund returned 10.96%, and trailed the Index by 2.05%.

On every rolling five-year period since we started the Fund, our returns (Institutional Class) have beaten the market 93% of the time. Over the last 12 years, our process has led to very consistent annualized returns for the Fund of 12.61%, which has outpaced the Index by 3.81% annually. And this is AFTER management and other fees. So definitionally, our investors have meaningfully outperformed small-cap index funds.

Of course, this strategy will not work in the short periods of time in which the market is fixated on “hype” versus objective long-term valuation metrics. This is what occurred in spectacular fashion in the third quarter of 2025, and to some extent in the fourth quarter as well. In these periods, low quality (high debt and poor profitability) and short-term price momentum-oriented stocks outperformed. Because low quality stocks outperformed, our portfolio experienced a 4.7% headwind versus the Index in 2025 just for owning higher quality stocks! This was the worst year on record for underperformance by high quality* stocks (going back to when data first started to be collected in 1975). We doubt it will re-occur. Similarly, short-term momentum* was a 2.8% headwind to the Fund in 2025 versus the Index due to our long-term orientation. Stacked together, these “factors” hurt the Fund’s relative performance versus the Index by about 7%, putting the Fund’s overall 2% relative performance shortfall into some context. We plead guilty to not owning quantum computing companies with minimal revenues until the 2030s (or small nuclear reactor companies with the same profile). Not owning some of these speculative areas hurt us (relatively). This has just been the nature of the stock market for the second half of 2025. History informs that this should almost certainly change.

* Based on MSCI Barra factor performance.

The top 10 positions in the Fund represented 26.7% of the net assets, and cash was 3.3%. Both of these were consistent with historical levels for the Fund.

We remain high quality, long-term, and “anti-momentum” investors (contrary to the style of the 2025 market). What do we mean by “anti-momentum?” Basically, we do not buy stocks just because they are going up (the definition of momentum investing). In fact, our process makes us more likely to buy high quality companies when their stock prices are down, valuations are compelling, and when we have a divergent view from the market. Classically these are “fallen angel” investments.

We added to our position in Exact Sciences Corporation in 2025 at average prices of about $51. Exact was trading at what we viewed as an extreme multiple discount because investors were growing concerned that its fecal-based colorectal cancer tests would be replaced by blood draw tests. It received a buyout offer in the fourth quarter by Abbott Laboratories for a price of $105 in cash (we believe Abbott saw the same terrific cash flow progression for Exact as we did!).

We bought more indie Semiconductor, Inc. in the second quarter at an average price of $1.82, down around 55% in the quarter due to concerns about the company’s ability to grow. It ended the year at $3.53. Similarly, we purchased SiTime Corporation in the first quarter at average prices of $178.63 when it was down 17% (general concerns about the semiconductor sector) – shares finished the year at over $353. In each case, we “anti-momentum” bought on conviction based on our in-depth fundamental research.

We believe that the economy is set up extremely well in 2026 given:(1) inflation is lowering (it peaked at 9.0% in June 2022 and has now annualized at 2.7% in December 2025, which is 0.3% lower than three months ago); (2) massive capital is being spent in the U.S., including the buildout of AI datacenters (it is estimated by Goldman Sachs that $400 billion will be spent on AI capital in 2025, nearly doubling from about $230 billion in 2024 - and this will grow to over $525 billion in 2026, up over 30%), plus projects associated with trade deals that will bring construction to the U.S.; (3) there is meaningful deregulation in the federal government; (4) GDP is growing at a nice clip (4.3% real growth in the third quarter according to the Bureau of Economic Analysis – which should accelerate in 2026 as people get their tax refunds based on recent tax cuts); (5) there is a reduction in federal spending which is starting to lead to a reduction in both the annual federal budget deficit and the overall level of federal debt; and (6) the Federal Reserve is at the start of an interest rate reduction cycle.

Federal spending has flattened and the budget deficit has narrowed from $1.83 trillion to $1.78 trillion (down 2.7% from government fiscal year ending September 30, 2024 to 2025) partly due to an over 5% reduction in federal outlays in the first half of 2025, and partly due to tariff revenue growing enormously. In 2025 through September, new tariffs have raised $124.5 billion, or $156.5 billion adjusting for accelerated purchases by importers when the tariffs were initially announced (Penn Wharton Budget Model, December 23, 2025 update). The tariffs do not seem to be driving inflation, as the Administration has backed off its initially extreme rates to move rates more parallel to its trading partners.

There are concerns about the consumer due to affordability issues, primarily as home values (shelter costs) remain elevated despite mortgage rates coming down. Medical care costs have also increased, as have utility costs. So even though gasoline costs are actually down, overall cost of living concerns have taken down a lot of our Consumer Discretionary names like Floor & Decor Holdings, Inc. and RH (Restoration Hardware) for the full year, and great restaurant companies like Wingstop Inc. and Texas Roadhouse, Inc. were down in the third quarter. Lower income consumers are being hurt by higher loan costs, leading to higher defaults, which is why used car price inflation is so much higher than new car inflation. We believe that these issues will get better, and the tailwinds to the economy mentioned above are extremely strong.

There are also big concerns about AI destroying competitive advantages within the software industry (an area in which the Fund has a large exposure). Aside from a few larger capitalization market leaders, software multiples have generally compressed through 2025. It is believed by many that new ways of coding applications using AI prompts (so-called “vibe coding”) will make it easy to copy existing software. But the devil is in the details, and we posit that the complexity of integrating software into large enterprises is not easily replicated. Enterprise software (particularly “systems software” that deals with the guts of networking, cybersecurity, identity protection and data protection) must deal with heterogeneous networks (on-premise servers, added to one or more cloud service providers), as well as with many different operating systems and types of hardware. Such complexity and security needs favor companies that have proven, over long periods of time, that their code structure is resilient and safe. The more complex the installation, the more protected the software company should be.

Software generally has what we view to be the best possible business model. Companies in the space sport high margins (typically 70% to 90% gross margins or more), have recurring subscription revenue, get paid up front for such subscriptions (so they increase cash flows the more they grow – so called negative working capital), and typically have large amounts of net cash on the balance sheet (negative leverage). And the acquisition environment is running hot. In 2025, multiple investments were bought away from us – Couchbase, Inc. (by private equity), CyberArk Software Ltd. (by cybersecurity software giant Palo Alto Networks Inc.), and Clearwater Analytics Holdings, Inc. (private equity). Given what we view as objectively low valuations, meaningful competitive advantages, and rapid growth in free cash flow, we are extremely excited about our holdings in Dynatrace, Inc. (20%), Varonis Systems, Inc. (23%), SentinelOne, Inc. (44%), Netskope, Inc. (50%), JFrog Ltd. (22%), and Waystar Holding Corp. (15%).

Shares of Exact Sciences Corporation contributed to performance. Exact is a molecular diagnostics company focused on the early detection of colorectal cancer. It is best known for its non-invasive colorectal cancer screening stool test called Cologuard. The big picture is that half of the 106 million adults in the U.S. recommended for colorectal cancer screening are not up to date, and we think the Cologuard test presents the best combination of non-invasiveness and accurate sensitivity/specificity as compared to more invasive colonoscopies and less accurate blood tests. Exact shares outperformed in 2025 as the restructuring of the commercial team drove Cologuard volume growth reacceleration, and shares further outperformed as Abbott Laboratories announced an agreement to acquire Exact Sciences for $23 billion in November. The deal represents a >50% premium over the prior closing price.

Shares of Establishment Labs Holdings Inc. contributed to performance. Establishment Labs sells next-generation Motiva breast implants that have meaningfully lower safety risks and aesthetic benefits compared to competitors. In particular, Motiva implants cause significantly less capsular contracture (where surrounding tissue squeezes the implant) and no known cancer risks. Motiva also comes in a more natural Ergonomix shape with a natural-feeling softer fill as well as potentially a smaller implant scar. The company's implants have captured significant share in many international markets, and the U.S. launch is now underway and progressing well, having already captured about 20% of the U.S. aesthetic breast augmentation market. Reconstruction approval (for cancer survivors) should occur in 2026 as well. We believe Establishment will capture substantial plurality share in the U.S. over the next few years, which will drive significant revenue growth, profitability (it will be free cash flow positive in 2026), and value creation.

Clearwater Analytics Holdings, Inc., a provider of portfolio accounting and reporting software, contributed to performance. The company reported solid third quarter 2025 earnings and raised 2025 guidance. Clearwater also announced that a consortium of private equity investors intends to acquire the company, which drove a sharp positive market reaction. We retain conviction and believe Clearwater has meaningful competitive advantages and the potential to compound revenue at attractive rates for several years. The company has an efficient business model that should drive 40%-plus adjusted EBITDA margins over time.

Shares of Varonis Systems, Inc., a cybersecurity vendor focused on classifying and protecting corporate data, detracted from performance during the quarter. The company’s core web-based Subscription as a Service (SaaS) business (76% of total revenues) was strong, growing over 100% year-over-year fueled by new customer wins and existing customers protecting more data stores. However, Varonis’s legacy on-premise commercial and Federal businesses experienced higher-than-expected churn towards the end of September, causing management to lower its overall forecasts for the year. Varonis announced that it would be sunsetting its on-premise software by the end of 2026. Management believes transitioning to single code base and revenue model will drive better overall security outcomes and lower total cost of ownership for its customers, while simplifying the sales motion and driving better efficiency for the company. While the remaining transition creates some uncertainty around 2026 revenue, we believe Varonis will emerge with accelerating growth, better profitability, and a more predictable model going forward.

Liberty Live Holdings, Inc. was a detractor during the period. Shares traded sharply lower following Live Nation's third quarter earnings release. During the quarter, concert segment earnings fell short of investor expectations, a divergence from a string of significant earnings beats in the business line. Management also lowered its outlook for ticketing in 2026 due to measures taken in response to regulatory action around ticket buying “bots” and costs associated with the FTC lawsuit. Stepping back from near-term results, Live Nation has been executing well on developing its owned venues and the long-term growth opportunity remains attractive. Strong category demand and organic investment in its footprint underpin steady double-digit earnings compounding. We continue to own the stock due to a favorable view of the Live Nation business and an attractively valued entry point at the Liberty holding company level.

Primo Brands Corporation was a detractor during the period following an earnings release that showed a decline in the company's home and office water delivery business. As part of the merger between Blue Triton and Primo Water, management identified opportunities to save costs through elimination of redundant manufacturing plants and distribution branches. Execution of this plan during the first half of 2025 led to unexpected service disruptions in the home and office delivery segment. The problems were caused by changes to 5-gallon jug availability and adoption of new technology systems used by drivers to manage their deliveries. These issues negatively impacted the logistics of highly calibrated routes and led to a temporary reduction in on-time, in-full delivery rates. The third quarter results were a continuation of these issues, however, the magnitude of service disruption and related impact on revenues was greater than expected. We expect the business to return to its normal rate of growth in the coming quarters and believe the stock will command a materially higher valuation once the current challenges are in the rear view.

We initiated a position in Waystar Holding Corp., which provides a full spectrum of revenue cycle management software to health care providers. The company services 30,000 clients, representing over 1 million distinct providers. And it processes 6 billion transactions per year, including over $1.8 trillion in annual gross claims which span about 50% of all U.S. patients. Waystar’s $1 billion-plus in estimated 2025 revenues are a 50/50 combination of subscription and volume-based revenue (we believe volume is a true secular grower as health care claims are constantly increasing).

The software smooths the claims process in three distinct areas: (1) Intake (insurance eligibility checks, cost estimation, prior-authorization approval and co-pay collection); (2) Clinical Visitation (documentation integrity, showing medical necessity, accuracy of documentation and coding); and (3) Post Care (clean claim submission, claim monitoring, patient payments, denial/appeal, and payor remittance). Waystar acquired its Clinical Visitation capabilities with the July 2025 acquisition of Iodine Software for $1.25 billion, completing what we believe is the industry’s only end-to-end claims automation solution. Historically, this has been a manually intensive process, but it has been highly automated by Waystar, with the help of AI. There are many examples of the value of Waystar’s software. Iodine can use AI to take a physician’s written or typed notes, scan them, and compare them to insurer databases for medical justification, accuracy and completeness, thereby reducing denied claims. Waystar’s software can do similar things with denied claims, using AI to automate the appeals process. It is believed that 80% of the market is still managed by more manually automated business process outsourcing firms. So, there should be a lot of market share yet to convert, in an ever-growing market.

The company will grow revenues organically in the low teens through 2031, with about 45% EBITDA margins (a measure of cash flow before financing and non-cash costs) and 29% free cash flow margins (real cash) growing to 31% or better, as the company reduces debt service costs (used to fund acquisitions and as a result of its legacy private equity heritage). Because it generates so much growing free cash flow, net leverage for the company (debt minus cash balances) will go from 3 times EBITDA at the end of 2025 to about zero in 2028. And the company only trades at about 15 times its 2026 EBITDA – and only 6 times our 2031 EBITDA estimates.

We recently initiated a position in Casella Waste Systems, Inc., the fifth largest waste management and recycling company in North America. Founded in Vermont in 1975 by John and Doug Casella, the company has been through multiple iterations owning a variety of businesses, but over the last decade has transformed itself into one of the fastest growing waste companies in the New England and Mid-Atlantic regions. The company has strong pricing power with 70% of the business focused on less competitive rural markets and 15% of revenue tied to high margin disposal/landfill revenue. This strong pricing power has resulted in mid-to-high single-digit organic top-line growth and mid-teens free cash flow growth. The team has augmented its organic revenue growth with over 80 acquisitions in the last 7 years, finding high quality bolt-on-type deals where Casella is able to substantially increase margins over time. We would expect this pace of M&A to continue for the foreseeable future.

We purchased Casella’s stock after the company meaningfully underperformed peers in 2025 due to issues integrating a newly acquired Mid-Atlantic acquisition. These integration challenges caused the company’s margins to come in lower than expected. We believe the company has largely rectified the issues and it is back on track to expand its margins in 2026. If we are correct, we would expect investors to reward the stock with a higher valuation multiple that is more in line with its historical average.

Lastly, Casella is the only remaining small sized, publicly traded solid waste player amid giants Waste Management, Republic Services, Waste Connections, and GFL Environmental. While not core to our investment thesis, it is clear Casella is a scarce asset that could at some point in the future be a prime acquisition target for one of the larger players.

During the quarter, we initiated a position in GCI Liberty, Inc. GCI is the leading broadband cable provider in Alaska. GCI generates revenue from broadband and wireless services. Roughly 97% of Alaskans live within GCI's network footprint and about 80% have access to high-speed tiers (2.5 Gbps+). We estimate that GCI has 70% to 80% market share of high-speed broadband in Alaska. For mobile services, GCI operates its own mobile network competing in an oligopoly with AT&T and Verizon. Revenues are evenly split between residential and business customers.

Alaska has unique population density characteristics that create a very high-cost environment that necessitates government funding. In addition, the competitive environment for Alaskan cable is more attractive than in the lower 48 states. Most notably, fiber over-build has had limited success given substantially higher costs and fixed wireless access has limitations due to the harsh elements in Alaska. GCI is completing an upgrade cycle in 2026 resulting in improved free cash flow generation in 2027 and beyond. We believe the current valuation offers a double-digit free cash flow yield.

However, our decision to invest in GCI extends beyond the attractive free cash flow characteristics. A guiding principle of the firm is "we invest in people" and GCI offers the opportunity to participate in one of the final acts of John Malone's legendary career. Mr. Malone and the Liberty Media team, with whom we've invested alongside for many years, have been clear about their intent to use the cash flow and borrowing capacity at GCI to buy other assets in the media sector. We assess that GCI will have at least $1 billiion of cash to serve as an equity check in future transactions. Management has a broad mandate and expects to complete a transaction in 2026.

We established a position in JFrog Ltd., a leading provider of software supply chain management tools that enable developers to securely build, manage, and release modern applications. The company’s core offering, Artifactory, serves as a universal hub for "binaries” which is the ones and zeros language that computers speak (and use to run applications), after converted (compiled) from human facing programming languages such as Python, JavaScript, C#, and many others. As organizations develop more complex applications that are increasingly dependent on open-source libraries (pre-fabricated chunks of code that can be shared among users around the world), the management of these binaries has become a critical operational requirement. JFrog addresses this challenge with a centralized solution that stores, tracks, and secures software binaries, facilitating faster development cycles and consistent deployments. The platform also extends into security with modules like JFrog Xray and Advanced Security, which continuously scan binaries to ensure only compliant, vulnerability-free software reaches production.

JFrog is widely recognized as the industry standard for binary management, serving a customer base of over 7,000 organizations. This includes 83% of the Fortune 100, the top 10 global technology firms, the 10 largest financial institutions, and 9 of the top 10 health care organizations. The company is gaining market share from smaller competitors in the binary category due to its breadth and depth of coverage—Artifactory supports over 30 different code package formats and programming languages (far more than competitors) while offering more efficient storage, deeper security context, and tighter integrations with other developer tools. By streamlining workflows and mitigating security risks, JFrog delivers high return on investment, reflected in its impressive 97% customer gross retention rate and healthy 118% net expansion rate.

The mission-critical nature of the JFrog platform has driven strong unit economics. Trailing-12-month free cash flow margins have more than doubled over the last two years, reaching 28% in the most recent quarter. Looking ahead, we believe JFrog can sustain durable growth as the acceleration of Generative AI creates demand for managing new binary types, such as Large Language Model artifacts. Furthermore, we expect average deal sizes to increase as customers migrate to the cloud—which typically provides a pricing uplift—and adopt premium modules like Advanced Security. We believe this combination of pricing power, margin expansion, and structural growth positions JFrog to generate substantial free cash flow and long-term shareholder value.

Fourth Quarter Sales

We sold the bulk of our position in Exact Sciences Corporation upon the announcement of its acquisition by Abbott Laboratories. We trimmed positions in three of our defense companies (Kratos Defense & Security Solutions, Inc., Karman Holdings Inc., and Mercury Systems, Inc.) due to valuations that had run up meaningfully in the second and third quarters on the heels of an increased defense budget (including allocations of billions of dollars for unmanned vehicles and the Golden Dome missile defense system which is still being fully formed), and increased foreign military sales. We still view defense as a core holding in the Fund.

Conclusion

While we missed outperforming the Index in 2025, we take great pride that our process led to terrific absolute returns for the Fund. We are extremely optimistic regarding the prospects for each holding in the Fund, and we look forward to a high growth, low inflation set-up for the 2026 economy. Thank you for investing in the Fund along with us. We hope that you and your families have a happy, healthy, and prosperous 2026.