Baron Global Opportunity Fund | Q1 2026

Baron Global Opportunity Fund® (the Fund) declined 4.8% (Institutional Shares) during the first quarter, compared to the 3.2% decline for the MSCI ACWI Index (the Index), and the 7.7% decline for the MSCI ACWI Growth Index, the Fund’s benchmarks.

After three consecutive years of 25%-plus gains, we were likely due for a breather. Entering the year, we thought the investing backdrop was reasonably balanced. On the one hand, the global economy was healthy and growth was accelerating. Advancements in AI were offering wondrous new opportunities. Monetary authorities, particularly the Fed in the U.S., were still in the midst of easing cycles, and most importantly, valuations, while no longer cheap, were not demanding, if like us, you have a longer time horizon. On the other hand… global geopolitical tensions remained high. The Middle East was getting hotter, Russia’s invasion of Ukraine, in its fourth year, continued unabated, tensions between China and Taiwan continued to be elevated, tariffs and trade wars persisted – the macro related uncertainty was likely going to be a headwind at some point. We have written in the past that while all bear markets, corrections, and pullbacks are different, at their core, they are driven by fear, uncertainty, and doubt. Fundamentally, markets depend on stability and predictability. Every time stability and predictability are threatened – markets pull back. The war with Iran certainly qualifies as an event that increased fear, uncertainty, and doubt and reduced stability and predictability. The price of oil spiked to as high as $150 per barrel (physical brent crude), treasury yields jumped up, the probability of further interest rate cuts went down, the range of possible negative outcomes expanded significantly – and the markets sold off.

History teaches that geopolitical events have minimal lasting impact on equity markets. Carson Group analyzed⁴ 40 major geopolitical and historical events spanning 85 years, from Germany invading France in 1940 to Iran attacking Israel in 2024 and found that the S&P 500 Index lost a median of just 0.9% in the first month following such events and gained a median of more than 7% over the following 12 months. RBC Wealth Management examined⁵ 20 major post-World War II military operations and found that the S&P 500 Index fell an average of 6% from the initial shock to the trough, but in 19 of 20 cases, the market returned to its pre-event level within an average of just 28 days. Moreover, in the 20th century, the Dow Jones Industrial Average advanced from 66 to 11,497 – an increase of 17,391% (not including dividends) despite four costly wars, a Great Depression, and countless recessions. Our default assumption is that pragmatism, rationality, and plain self-interest will continue to prevail. We admit to simply playing the odds here, but in an unlikely event we are wrong, beating an Index will be the least of our problems.

From a performance attribution perspective, stock selection contributed 53bps to relative returns, while the sector allocation detracted 213bps. Though our overweight to the poorly performing Information Technology (IT) and Consumer Discretionary sectors hurt, not having any investments in Energy (+33.6%), Utilities (+8.4%), Materials (+6.6%), and Consumer Staples (+3.3%) cost the Fund 219bps, accounting for the entirety of the quarterly shortfall. From a geographical perspective, our 8.6% average underweight in the U.S. finally stopped hurting and actually added 22bps to relative results. The overweight to Brazil, the Netherlands, Taiwan, and Korea contributed 183bps, while the overweight to India and Argentina, and lack of investments in Japan and the U.K., detracted 164bps. Stock selection in the U.S., China, and Poland was strong, which was more than offset by poor stock selection in Canada, Korea, Brazil, and the Netherlands.

From a stock specific perspective, we ended the quarter with 9 contributors against 30 detractors. Our biggest contributor SpaceX, whose new funding round led our fair value committee to revalue the stock higher, which contributed 419bps to our absolute returns. We are very excited about this investment and believe that SpaceX continues to offer asymmetric, positive optionality as it gets closer to becoming a public company. ASML, Taiwan Semiconductor (TSMC), and InPost contributed over 50bps each, while GDS, Cloudflare, Codere, and BillionToOne added 59bps combined. On the other side of the ledger, 10 of our holdings detracted at least 50bps from absolute results due to broader AI and macro related concerns causing a sell-off. We do not believe any of these 10 investments will cause a permanent loss of capital and have in fact added to all of them, except for NVIDIA where we are maxed out due to a regulatory restriction.
 

In News Reporting – Sensational Always Dominates Over Statistical

The current news cycle often has a disproportionate impact on stock prices. Entire investment strategies are built on analyzing daily news flow and executing trades with lightning speed (think micro-seconds). An army of technical analysts opine daily on the significance of “breaking moving averages” and formations of “head and shoulders” in various stock charts. The goal of these highly paid experts is to predict where stock prices will go over the next say 30 or 60 days. Most of our investors will agree with us that 30 or 60-day returns don’t matter, though they can easily make us lose sight of our long-term time horizon. We do not make investment decisions based on timing or predicting the outcomes of wars or the price of oil. Not only because macro-driven events are notoriously difficult to predict with any consistency, but more so because we believe they will work themselves out. So, with rationality and pragmatism prevailing as our default setting, we focus on separating the signal from the noise and identifying the statistically significant events that help us build conviction in the investment thesis for a long-term success of a business.
 

“AI has moved from being a novelty to becoming something that is really useful! Inferencing is exploding and we are standing smack in the middle of this tornado.”
– Andrew Feldman, CEO of Cerebras⁶

As we’ve been doing for years now, we took our semi-annual Baron technology trip to Silicon Valley in March, meeting with venture capitalists, and companies across private and public markets and across software and semiconductors, both owned and not owned. Over the last three years, after every trip, we came away thinking that AI disruption is real and is likely to be paradigm shifting. This time, it felt different. AI was everywhere. It dominated conversations and was a key item on everyone’s agenda. During the weekend before we got there, Andrej Karpathy, a leading AI authority, released AutoResearch, a project that allows AI agents to autonomously run research loops, modify code, run training experiments, and evaluate results – essentially AI training AI⁷ . A couple of days later Base44 (owned by Wix), released Superagents, which enables users to build AI agents that can enhance workflows, and we found ourselves at 11pm one night building an agent that sends us daily email summaries with all the important news items for the companies we own.

One of the main investor concerns over the last six months has been that the mammoth increases in capex and capacity buildouts were not accompanied by the clear evidence of revenue generation. This is no longer the case. The most significant development in the quarter, in our opinion, was the dramatic acceleration in AI adoption and usage. Witness the curious case of Anthropic, the maker of Claude Code and Claude Cowork, which reported annualized recurring revenues (ARR) of $9 billion at the end of 2025, projecting to reach $30 billion in ARR exiting 2026. The company then proceeded to add $4 billion in ARR in January, $6 billion in February, and a mind-boggling $11 billion in five weeks through the first week of April⁸ , surpassing the $30 billion target for 2026. Anthropic added over $21 billion in net new ARR in just over one quarter. Think about that for a minute… Dario Amadei, the company’s CEO, disclosed a few months ago that a majority of revenues is coming from enterprise customers with over 1,000 companies paying over $1 million in ARR. These are the most coveted customers in the world that every business dreams of getting. When will AI revenue finally show up? Well… it just did! While clearly the most impressive, Anthropic is not alone. OpenAI’s ARR surpassed $25 billion⁹ from $20 billion at the end of 2025. As of the writing of this letter, no public company has reported its first quarter results, but we did get a glimpse from Amazon’s CEO, Andy Jassy, who released his 2025 Letter to Shareholders on April 9¹⁰. Amazon AI is now at a $15 billion revenue run rate (260x larger than original Amazon Web Services (AWS) at the same point), their combined custom chips business (Graviton, Trainium, and Nitro) is now at a $20 billion ARR, growing triple-digit percentages year-over-year, implying approximately 13% of total AWS revenues. And that number is understated since they are monetizing the chips themselves. If it was a standalone business selling to third parties, their ARR would be $50 billion. AWS continues to be capacity constrained – growth would be higher if they could serve it. At 5.0% of the Fund’s net assets, Amazon is our fourth largest holding – we would own more if we could.
 

Culture, Structural Moats, Adaptability to Change, & What’s Not Going to Change

“I very frequently get the question: 'What's going to change in the next 10 years?' And that is a very interesting question; it's a very common one. I almost never get the question: 'What's not going to change in the next 10 years?' And I submit to you that that second question is actually the more important of the two — because you can build a business strategy around the things that are stable in time."
– Jeff Bezos¹¹

We have often used this quote when analyzing and thinking deeply about the companies we invest in. What attributes of successful businesses will not change in the world of AI?

• Solving real problems for customers.
• Doing it in a way that is unique – with competitive advantages that are durable.
• Being adaptable to change – see our 4Q2025 letter for an in-depth discussion.
• Management that thinks and acts like owners of the business, optimizing for the long term.
• Management that is willing to make big bets when inflection points (a disruption) become clear.

Here too, Amazon’s shareholder letter is instructive – addressing all of these points. Jassy begins with what won't change: "We believe that customers will always care deeply about massive selection, low prices, very fast delivery, ease of use, and how they're treated." But he argues that AI will reshape how those needs are served, and that leaders must resist the temptation to simply bolt AI onto existing experiences: "The trick for leaders, ourselves included, is how to get organized and convicted about going back to the starting line and reimagining your experiences from a clean sheet of paper." If this sounds familiar, it is because this is exactly how Elon Musk has described his approach to innovation. It is all about first principles, reimagining the entire experience, disregarding the “this is how we’ve always done this” mentality and thinking.

Jassy then describes what adaptable culture looks like in practice – when Amazon's Bedrock team realized it needed an entirely new inference engine, rather than patching the existing one, it spun out a group of six engineers who rebuilt the architecture from scratch in 76 days using Amazon's own AI coding tool, Kiro. The result – an engine called Mantle – became the backbone of Bedrock, which nearly doubled month-over-month in March and processed more tokens in Q1 2026 than in all prior years combined. The right culture and organizational structure are preconditions for this kind of adaptability. "You need to move fast, have teammates that act like true owners, and be scrappy. At Amazon, we talk a lot about operating like the world's biggest startup. It's the primary reason we've worked to flatten our organization last year." Shopify CEO, Tobi Lütke, has taken adaptability further than perhaps any leader in our portfolio: in April 2025 he published an internal memo¹² declaring "Reflexive AI usage is now a baseline expectation at Shopify," mandating that before requesting new headcount, teams must demonstrate why AI cannot do the job.

Many of our most successful investments share one attribute: founder-led management that thinks and acts like long-term owners. Jensen Huang has led NVIDIA for 33 years since founding it in a Denny's in 1993. Tobi Lütke has led Shopify for over two decades. Jeff Bezos wrote in his original 1997 shareholder letter – which Amazon still appends to every annual report as a statement of enduring values – "We will continue to focus on hiring and retaining versatile and talented employees... each of whom must think like, and therefore must actually be, an owner."

And when the inflection is big enough, Jassy argues, you must bet disproportionately. Amazon guided 2026 capex to $200 billion. "If you believe you've found one of these disproportionate shifts, you want to invest as aggressively as you responsibly can. This will create investment spikes that will invite scrutiny, but the game-changers don't typically accommodate smoother investment horizons… Inflections aren't usually smooth or calm. They favor the bold and adaptable."
 

What does AI mean for the durability of competitive moats?

Does AI accelerate disruptive change? Yes. Does it reduce barriers to entry? Yes. Does it make it easier to copy features and functionality from competitors? Yes. Does it mean that competitive advantages are no longer sustainable? Not in our view. Instead, we would argue that AI changes the nature of competition, increasing the importance of structural competitive moats while reducing the durability of simple, product/feature/workflow-based moats. Some of the structural competitive moats that in our view remain durable include the following:

• Platform businesses with network effects – Amazon is the poster child for network effects where a high number of loyal consumers (repeat buyers) attracts the highest numbers of merchants who offer the widest variety of products at very competitive prices, which attracts more loyal consumers and so on. Cloudflare built a network that operates in over 330 cities across 125 countries where it cross-connects its customers directly with origin servers, bypassing the public internet, significantly reducing latency and reducing costs. The more traffic that flows through Cloudflare, the better its routing intelligence becomes, making it more attractive for the next customer. AI further reinforces this moat as its close physical proximity to users globally enables it to offer low-latency AI inference, which in turn drives demand for its core offerings. Matthew Prince, Cloudflare’s CEO articulated it on a recent earnings call: “This creates a virtuous flywheel: more agents drive more code to Cloudflare Workers, which fuels demand for our performance, security, and networking service. ” Shopify’s network effects are multi-faceted – the millions of merchants on its platform with real time transaction-level data enables it to innovate rapidly, having a front row seat to what makes merchants successful. That same data powers a virtuous cycle across the business: it fuels AI Sidekick, which uses aggregated insights to suggest actions that help merchants improve their operations; it enables Shopify Capital to underwrite loans to merchants with superior risk models. Shopify’s scale on the consumer side (over 150 million consumers use Shop App and Shop Pay) and consumer trust drives higher conversion rates for Shop Pay (its instant purchase button), making Shopify more attractive to prospective merchants. In agentic commerce, the two-sided merchant-consumer network effect compounds further: as more merchants join, the product catalog becomes richer, making it the most attractive ‘rails’ for AI chatbots to access the widest selection of products, which in turn attracts more merchants to Shopify. Harley Finkelstein on a recent earnings call: “No one, and I mean no one, understands this like Shopify… This is a transitional moment in Shopify's history. We are now designing the new normal, just like Tobi predicted a decade ago, and it will fundamentally change our position in the world.”
   
• Proprietary data moats – In the age of AI, continuously generated proprietary data that enables ongoing product improvement is an important moat. It cannot be replicated from public sources by foundation models. CrowdStrike processes trillions of endpoint security events globally. CEO George Kurtz, described it on a recent earnings call: “Our data moat creates a structural advantage no LLM provider can replicate.” Snowflake’s Cortex Code (CoCo), unlike generic coding agents, operates natively inside Snowflake with deep awareness of each customer's schemas, permissions, governance rules, and data relationships – context that no external AI tool can access. As CEO Sridhar Ramaswamy explained: "It understands Snowflake intimately, all its different capabilities. We are seeing productivity gains of 5 to10 times." ¹³ Each interaction makes CoCo smarter about that customer's environment, creating a feedback loop that widens the moat over time, by increasing the value of Snowflake for its customers. Bajaj Finance illustrates how proprietary data compounds into product advantage in financial services. The company has lending data on over 100 million Indian consumers across 26 product categories. In the last quarter alone, its AI systems listened to 20 million customer calls, converted voice to text, extracted structured data on 520,000 customers, and generated 100,000 new personalized loan offers that the company previously lacked the information to make, producing ₹1,600 crore in disbursals, roughly 10% of quarterly volume. This is a capability that did not exist two quarters earlier. Each call makes the models smarter: better at identifying which customers to target, what products to offer, and when to offer them. Vice Chairman Rajeev Jain commented on the earnings call: "We are not testing AI. We are deploying AI across the board, across the life cycle."
   
• Economies of scale – MercadoLibre is a great example of how scale creates a self-reinforcing cycle. Greater volume drives lower shipping costs per package, enabling the company to reduce its free shipping threshold, which drives higher purchase frequency, expands the addressable market to lower-cost everyday items, and attracts more buyers and sellers, which further improves logistics density, reducing unit shipping costs and so on. In the Summer of 2025, MercadoLibre lowered its free shipping threshold in Brazil and the results were striking: items sold accelerated from 26% growth in Q2 to 42% in Q3 and 45% in Q4. Unique buyers crossed 80 million for the first time, and the company achieved record market share gains in both Brazil and Mexico – all while unit shipping costs in fulfillment fell year-over-year. CEO Ariel Szarfsztejn on a recent earnings call: "Growth has accelerated, frequency has accelerated. We had record conversion rates, record retention rates for new and existing buyers. We are expanding market share and reaching record levels." Nearly 75% of fast shipments are now delivered within two days – services possible only at MercadoLibre's scale. Of course, Amazon’s scale in the U.S. (and other international markets in which it is a leader) similarly provides it with the density needed to have fast delivery at low-per-unit shipping costs that are hard for competitors to replicate. Coupang’s scale enabled it to build a fulfillment infrastructure so dense that 70%-plus of South Koreans live within seven miles of a fulfillment center, a physical network no AI model can shortcut. Its scale also enables it to offer a wider selection of products at better prices to customers, which makes its offering unmatched, with its competitive moat only widening as it scales further.
   
• Manufacturing complexity and accumulated know-how – TSMC manufactures approximately 90% of the world’s leading-edge semiconductors. Its moat is structural: building a cutting-edge fabrication facility costs over $20 billion and takes many years, but the real barrier is yield – the percentage of functional chips per wafer. TSMC has spent decades perfecting yields at each successive process node through continued manufacturing iterations, accumulating institutional knowledge over time that is proprietary and can NOT be replicated by AI models. Acceptable (i.e. high) yield is becoming increasingly more important and harder to achieve as chip complexity and prices increase over time. While Samsung and Intel also spend large amounts of capital building advanced fabs, they serve a fraction of TSMC’s customer base, which means they cannot amortize fixed costs as efficiently and cannot accumulate yield-learning data as quickly. This creates a self-reinforcing cycle: higher yields attract more customers for TSMC, which enables it to earn higher cross-cycle returns on its investment (while the diversity of its customers also reduces the overall cyclicality of the business), which funds more R&D, which enables TSMC to reach the next node quicker and with higher yields than competitors, which widens the lead further. Advanced nodes (7 nanometers (nm) and below) now generate 74% of TSMC’s wafer revenue. ASML also occupies an enviable competitive position: it is the sole manufacturer of extreme ultraviolet (EUV) lithography machines, the equipment required to “print” transistors at the most advanced nodes. There is no alternative supplier. Each machine costs approximately $350 million, takes over a year to build, is extremely complex, and combines cutting edge technologies from a whole ecosystem of global supply chain from Cymer’s lasers that shoot laser pulses at a rate of 50,000 times per second at liquid tin droplets, to Zeiss’ optics with mirrors that boast sub-atomic level smoothness.
   
• Regulatory moats and switching costs – argenx holds FDA approval for Vyvgart – clinical data, patents, and physician trust cannot be replicated by an AI algorithm. Heartflow accumulated over 600 peer-reviewed clinical papers, has FDA approval for its heart-disease diagnostic from the FDA, and also has over 600 patents. Back to TSMC, rising chip complexity increases the switching costs for customers such as Apple, NVIDIA, Advanced Micro Devices (AMD), and Broadcom, as changing suppliers would take years, add major costs, and risk product delays (which could in turn negatively impact their competitive positioning). It’s not about whether a customer can switch, but about whether it’s in their best interest to do so.

We have decades of experience in identifying and investing in businesses that are beneficiaries of disruptive change. While the AI disruption feels like the most significant and challenging disruption of our careers (and frankly, our lives) we have built internal processes that we believe have prepared us well for navigating it. We have learned to think probabilistically (expected value = every possible outcome multiplied by the probability of each outcome happening) and believe there is an advantage in allocating capital against the entire range of outcomes rather than the best, the worst or even the most likely outcome. We guard against behavioral biases and systematically seek out disconfirming evidence to stress-test and re-underwrite the key assumptions that form an investment thesis for every company that we own. The goal is to reduce investments in businesses where our conviction level has lessened and increase investments where our conviction level has increased.

We continue to rehabilitate the Fund from challenging performance returns in 2021 and 2022. Over the last three years, the Fund has returned 20.8% per year, net of all fees and expenses, which compares favorably to 16.6% for the Index, and 12.6% for the Morningstar Global Large-Stock Growth Category Average (Peer Group), ranking the Fund in the 8^th percentile. The longer-term, 10-year performance is also now strong with a 14.0% annualized return, outperforming the Index by 264bps annually and ranking in the 7^th percentile of the Peer Group.*

Space Exploration Technologies Corp. (SpaceX) is a high-profile private company founded by Elon Musk. The company's primary focus is on developing and launching advanced rockets, satellites, and spacecrafts, with the ambitious long-term goal of making life multi-planetary. SpaceX is generating significant value with the rapid expansion of its Starlink broadband service. The company is successfully deploying a vast constellation of Starlink satellites in Earth's orbit, reporting substantial growth in active users, and regularly deploying new and more efficient hardware technology. Furthermore, SpaceX has established itself as a leading launch provider by offering highly reliable and cost-effective launches, leveraging the company's reusable launch technology. SpaceX capabilities extend to strategic services such as human spaceflight missions. Moreover, SpaceX is making tremendous progress on its newest rocket, Starship, which is the largest, most powerful rocket ever flown. This next-generation vehicle represents a significant leap forward in reusability and space exploration capabilities. We value SpaceX using prices of recent financing transactions.

Dutch semiconductor equipment company ASML Holding N.V. shares increased 23.5% in the first quarter due to robust demand for its lithography systems amid a strong AI-driven semiconductor capex cycle. ASML holds a monopoly on extreme ultraviolet (EUV) lithography, the indispensable technology required to manufacture the world's most advanced chips at 7nm and below. Without EUV, chipmakers cannot economically achieve the transistor densities needed to power AI accelerators, flagship smartphones, and autonomous vehicles. As leading chipmakers including Taiwan Semiconductor (TSMC), Samsung, and Micron race to expand advanced manufacturing capacity to meet surging AI demand, ASML sits at the center of the global semiconductor ecosystem as an indispensable enabler. Moreover, we expect a strong product cycle over the next five years as High-NA EUV – the next-generation platform delivering superior resolution and continued transistor scaling – enters high-volume manufacturing. We also project significant gross margin expansion, driven by ASML's pricing power and increasing scale, supporting strong double-digit earnings growth.

Semiconductor giant Taiwan Semiconductor Manufacturing Company Limited shares rose 11.4% in the quarter, as revenue growth of 20.5% (and 25.5% in USD) exceeded expectations due to surging demand for AI chips. TSMC dominates the advanced semiconductor foundry market, controlling over 90% share of cutting-edge sub-7 nm nodes that power AI servers, flagship smartphones, and autonomous vehicles. The company benefits from a virtuous cycle in which its massive scale and profitability generate the capital necessary to fund industry-leading R&D and capex, in turn widening its technological moat and reinforcing its pricing power. As the ultimate picks-and-shovels provider of the AI era, TSMC remains insulated from the competitive dynamics of the AI chip design ecosystem. Whether hyperscalers develop custom accelerators or deploy merchant GPUs from companies like NVIDIA and AMD, nearly all advanced AI accelerators are manufactured exclusively at TSMC’s 3nm and 5nm nodes. We believe TSMC will deliver 20% earnings growth over the next several years, supported by secular AI-driven demand for leading-edge manufacturing capacity.

Shares of Shopify Inc., the leading global commerce operating system, declined 25.5% during the first quarter, as a broader software selloff compressed high-multiple growth stocks. The business itself delivered an exceptional year: gross merchandise value (GMV) grew 30% - adding $86 billion of incremental volume - revenue grew 30%, and free cash flow grew 26% to over $2 billion. Management guided to low-thirties revenue growth for Q1 2026, above the roughly 25% most analysts had modeled. Q1 FCF margin guidance of low-to-mid- teens reflects deliberate continued reinvestment in the business to take advantage of the large opportunity ahead. Enterprise penetration remains in early innings: merchants above $25 million GMV are growing faster than the broader business, and Shopify holds low single-digit share of the $2.4 trillion enterprise e-commerce market. Shop Pay GMV of $121 billion grew 62% and now exceeds 50% of U.S. payments volume - a consumer trust signal with advertising optionality via the Shopify Product Network. We view the drawdown as an opportunity to own an exceptional business at a more attractive entry point, added to the position, and see a path for attractive multi-year compounding from current prices.

Snowflake Inc. is a cloud data platform primarily used for data analytics. Shares declined 31.3% during the quarter, as investors became increasingly concerned that AI-native competitors could disrupt traditional software vendors. Even so, the company posted solid results, with product revenue growth of approximately 30%, best-in-class 125% net revenue retention, record new customer wins, and robust cash generation. Backlog continued to build momentum, with accelerating growth as the company inked its largest contract ever, valued at over $400 million. Snowflake’s product innovation also remains strong, as the company rapidly embeds AI capabilities across its platform to better address customer needs. Its AI product suite has become the fastest-growing offering in the company’s history, already generating over $100 million in annualized revenue, with newer tools such as Snowflake Intelligence and Cortex Code now being used by thousands of customers. These developments support our long-term conviction in Snowflake’s ability to sustain durable growth and maintain its position as a foundational data platform, even if near-term valuation pressures persist.

Bajaj Finance Limited is a leading non-bank financial company in India. Shares declined 22.3% during the quarter as geopolitical tensions over the past month raised expectations of higher inflation and disrupted India’s easing interest rate environment, which could negatively impact consumption-led credit growth in the short term. We retain conviction in the company due to its best-in-class management team, robust long-term growth outlook, and conservative risk management frameworks. We believe Bajaj is well positioned to benefit from growing demand for consumer financial services in India, including mortgages, personal loans, credit cards, and other related products. We also believe the company will be an AI beneficiary thanks to its robust proprietary data and tech stack, which enables it to both run the business more efficiently and drive incremental loan growth.

The portfolio is constructed on a bottom-up basis with the quality of ideas and conviction level playing the most significant role in determining the size of each individual investment. Sector and country weights are an outcome of the stock selection process and are not meant to indicate a positive or a negative view.

As of March 31, 2026, the top 10 positions represented 60.9% of the Fund’s net assets, and the top 20 represented 82.9%. We ended the first quarter with 43 investments. Note that our top 30 investments represented 93.7% of the Fund.

Our investments in the IT, Industrials, Consumer Discretionary, Financials, and Health Care sectors, as classified by GICS, represented 97.8% of the Fund’s net assets. Our investments in non-U.S. companies represented 42.2% of net assets, and our investments in emerging markets and other non-developed countries (Argentina) totaled 27.3% of net assets.

* Individual weights may not sum to the displayed total due to rounding.

During the first quarter, we initiated two new positions – the electrical distribution equipment manufacturer, Forgent, and the AI cloud, Nebius.

We also took advantage of inflows into the Fund and stock price volatility to add to 22 existing investments including to Amazon, whose stock sold off to an all-time-low valuation, Shopify, whose stock declined over 25% despite the company’s solid fundamental execution, and Coupang, whose stock sold off due to a cyber-security incident, whose impact we believe will be short-lived. We also took advantage of AI fears to add to several of our highest conviction software names including: the leading cyber-security platform, CrowdStrike, the leading networking and cyber-security solutions provider, Cloudflare, the zero-trust networking provider, Netskope, the observability platform, Datadog, and the data platform, Snowflake. We also added to our aerospace and defense parts supplier, Loar, whose stock sold off on the potential short-term headwind of rising energy prices to travel; to our Chinese data center company, GDS; and to Nu, the leading Latin American neobank, after its stock sold off on an apparent miss in loan loss provisions - one driven entirely by accounting rules requiring front-loaded provisioning on expanded credit limits, not any deterioration in underlying credit quality.

Our largest new add in the quarter was Forgent Power Solutions, Inc., a leading manufacturer of electrical distribution equipment used in data centers, power grid, and energy-intensive industrial applications – transformers, switchgear, power distribution units, and prefabricated power skids. Forgent’s products are mission critical with stringent safety and reliability requirements focused on the most technically demanding applications – data centers, semiconductor fabs, power plants, battery energy storage, pharmaceutical production – with a high percentage of engineered content, tight tolerances, and rigorous certifications, requiring specialized labor with significant expertise. 85% of Forgent’s revenues are derived from the data center, grid and industrial end markets, which benefit from secular tailwinds such as the AI buildout, the growth in electrical load and the reshoring of industrial capacity into the U.S. These tailwinds have driven rapid above-industry growth with a 69% year-on-year revenue growth in the December quarter. Forgent is a low- and medium-voltage equipment specialist and focuses on custom, “engineered-to-order” products (which represent over 90% of its revenue), whereas larger competitors in the industry generally focus more on higher voltage and standard products. This segment of the market accounts for about a quarter of the market but is growing faster at a 25% CAGR (compared to high teens growth for the overall market).

The company differentiates itself from competitors by engaging deeply with customers in the design phase, offering custom products, which would typically take longer to manufacture, in shorter lead times than standard products, thanks to its unique customer engagement model, manufacturing floorplan layout, and supply chain management – which also enable it to earn attractive margins. The company also has an experienced management team and has invested in manufacturing capacity to support $5 billion in revenue, which, once completed, will give it one of the largest state-of-the-art manufacturing footprints in the industry. The company’s backlog of around $1.5 billion and robust recent orders, underpin strong revenue growth visibility. Despite inefficiencies from excess capacity, Forgent already has close to best-in-class EBITDA margins, which we expect to continue expanding as the company drives more volume over its manufacturing footprint. To date, most of its data center business has focused on colocation customers and neoclouds, with new, large opportunities to engage with hyperscale customers going forward.

We also initiated a small investment in Nebius Group N.V., a builder of a new AI cloud business, popularly referred to as a “neocloud.” Nebius is a Big Idea with a remarkable origin story – the company was born from Yandex, popularly known as the “Google of Russia,” which founder Arkady Volozh built into a $30 billion business over 25 years with leading positions in online search, e-commerce, ride-hailing, music streaming, maps, and cloud services. After Russia’s invasion of Ukraine, Volozh divested all Russian assets in the largest corporate exit from Russia in history ($5.4 billion), reconstituting the company as Amsterdam based, Nebius. The company boasts a world-class 1,300-strong team of engineers with decades of experience building large-scale computing systems and a vision to build a leading AI cloud business from first principles – a purpose-built vertically-integrated hardware and software stack optimized for AI workloads.

Nebius’ long-term vision requires significant resources to build the physical infrastructure and acquire customers. In the interim, Nebius is strategically and very selectively signing bare-metal GPU deals (renting the data center with the GPUs installed but no software on top of the GPUs) with Microsoft (up to $19 billion) and Meta (up to $27 billion). While there is a range of outcomes on the long-term value of GPUs in a bare-metal model (with the main concern revolving around the rapid depreciation of old GPUs as new more efficient ones are introduced), the useful life of AI accelerators appears to be longer than previously anticipated, as the economic output of GPUs (as measured by token throughput) has been increasing over time as models have improved, meaning project returns get better with age. Dylan Patel of SemiAnalysis explained the dynamic on the Dwarkesh Podcast (March 2026)¹⁴: GPT-5.4 from OpenAI generates more tokens on H100s than GPT-4 did, despite being a far more capable model, because newer architectures (such as sparse mixture-of-experts) are more computationally efficient per token. The implication is that an H100 is worth more today than when it was purchased three years ago. For Nebius, this means the residual value of its GPU fleet after long-term contracts expire could be substantially higher than depreciation schedules assume. The rapid growth in AI-demand has also driven H100 one-year rental prices higher by approximately 40% from $1.70 per hour in October 2025 to $2.35 per hour by March 2026 (SemiAnalysis, March 2026),¹⁵ even as NVIDIA’s newer Blackwell GPUs entered the market. In our view, as long as incremental AI token demand exceeds the token supply enabled by the new chips produced in a given year, older GPUs retain – and can even gain – value (as long as token throughput increases over time as models improve – despite an increase in intelligence). More importantly, these deals function as a quasi-financing mechanism for the AI cloud buildout. Meta's deal in-particular provides access to investment grade borrowing costs with no equity dilution, while acting as a backstop customer if enterprise demand for the AI cloud doesn't materialize on schedule.

Today we value the business from three angles: as a bare metal provider to technology giants like Microsoft and Meta, as an AI cloud business, and through its ownership of multiple high-growth companies. We agree with management's view that the AI cloud business is the big prize worth playing for, while the bare metal business is more of a funding mechanism to support the larger AI cloud vision. With that said, just the existing bare metal deals are worth mid-single digit billions in equity value (and Nebius may continue signing more of these deals, which earn a solid 20%-plus operating margin and which could have longer-term value optionality for GPUs as the deals lapse). Nebius' ownership stakes in ClickHouse, Avride, and Toloka reflect an additional high-single-digit billions in value (for example, ClickHouse, in which Nebius holds nearly 30%, is a fast-growing data platform, benefiting from the rapidly expanding demand for real-time data analytics driven by AI; it was valued at $15 billion in its last funding round from early 2026). This implies the market currently values the AI cloud business itself at roughly mid-teens billions of dollars, a figure that is justified if the company builds just a low single-digit billion dollars of AI cloud revenue over the next five years, which is less than 1% of market share assuming AI cloud total addressable market exceeds a trillion dollars by the end of the decade. We believe Nebius can do meaningfully better than that, taking share from the big three incumbents and capturing more than a few percentage points of this large addressable market, with the company targeting approximately 10%. At today's price, we are essentially buying a contracted bare metal business and a portfolio of valuable stakes at a fair price, and getting what could become a very big idea, as one of the world's next great AI cloud platforms with a world-class Founder/CEO and a team of engineers who have succeeded before at an attractive entry point.

We have conviction in Nebius' ability to build a large AI cloud business. They are leading the neocloud space in building a full suite of software offerings on their platform, much in line with what the big three hyperscalers have built for cloud workloads, i.e., multi-tenant compute, unified storage, inference-as-a-service, and security certifications, the kind of platform depth that took incumbents years to assemble. Key customers like Shopify, Cloudflare, and Revolut validate the technical prowess of their platform for AI workloads. Arkady Volozh, Founder and CEO, has spoken about Nebius ambition¹⁶:

“So we started with 25 megawatts, 18 months ago. Today, we're running 10x of that, almost. We publicly announced that we will be running another 5x on that. It will be almost 1 gigawatt of active power this year… We have a pipeline already reserved for 2 gigawatts, and it will be probably close to 3 gigawatts this year... But what we are building is not just gigawatts. We're building a real platform for AI developers, a real cloud, full stack cloud… This is a new cloud with a lot of services on top of it. It's not just multi-tenant cloud, which means basic cloud services, multi-tenancy billing, security, storage, managed databases, it's all there. But on top of it, there is a lot of services for – specific for AI builders like our Token Factory, inferencing platform or agentic search services, which people who build AI need, or just recently, we announced human experts in the loop there…”

“You manifest a future so convincing that there is no way it won’t happen – and then you endure the suffering in between.” 
Jensen Huang (NVIDIA’s Co-Founder and CEO) on Lex Fridman podcast, March¹⁷

Jensen’s philosophy captures perfectly how we think about investing in disruptive change. That is what investing in Big Ideas requires. It is never a straight line. There is always suffering in between. There are drawdowns, counter-narratives, and events that test your conviction and make you lose sleep at night. Patience, ability to separate the signal from the noise, and courage of conviction are needed to stay the course. While the talking heads are fixated on the price of oil or the latest posts on Truth Social, we believe we are on the cusp of something extraordinary. Jensen revealed a $1 trillion order book and proclaimed that OpenClaw “did for agentic systems what ChatGPT did for generative systems.”¹⁸ Dario Amodei from Anthropic said we are “near the end of the exponential.”¹⁹ Elon Musk called AI “the supersonic tsunami.”¹⁸ Across our portfolio – from Snowflake to CrowdStrike, Cloudflare to Shopify, Amazon to Datadog – the companies we own are reporting positive early data points on how AI is likely to be transformational for their businesses.

Every day we live and invest in an uncertain world. Well-known conditions and widely anticipated events, such as Federal Reserve rate changes, ongoing trade disputes, government shutdowns, and the unpredictable behavior of important politicians the world over, are shrugged off by the financial markets one day and seem to drive them up or down the next. We often find it difficult to know why market participants do what they do over the short term. The constant challenges we face are real and serious, with clearly uncertain outcomes. History would suggest that most will prove passing or manageable. The business of capital allocation (or investing) is the business of taking risk, managing the uncertainty, and taking advantage of the long-term opportunities that those risks and uncertainties create.

We are optimistic about the long-term prospects of the companies in which we are invested and continue to search for new ideas and investment opportunities while remaining patient and investing only when we believe the target companies are trading at attractive prices relative to their intrinsic values.

We appreciate your partnership in this journey.

Sincerely,