Baron Opportunity Fund | Q2 2026

During the second quarter, Baron Opportunity Fund® (the Fund) surged 27.07% (Institutional Shares), outperforming the Russell 3000 Growth Index (the Benchmark), which gained 17.05%, and the S&P 500 Index, which rose 15.20%. For the first six months of 2026, the Fund ascended 15.79%, beating the Benchmark, which appreciated 5.88%, and the S&P 500 Index, which advanced 10.21%.

Market Backdrop

After selling off in March amid the U.S.-Iran conflict and disruptions in the Strait of Hormuz, U.S. equity markets rallied to record highs during the second quarter, with most gains concentrated in April and May. The AI secular growth narrative was the principal driver of returns, overshadowing uncertainty surrounding a potential resolution to the U.S.-Iran conflict, shifting Federal Reserve rate expectations, and depressed consumer sentiment stemming from persistent inflation, high living costs, and elevated borrowing rates.

Market leadership was quite narrow during the quarter, with a dozen securities accounting for most of the S&P 500 Index’s gains. The 12 largest contributors were either Information Technology (IT) stocks or members of the Magnificent Seven complex—companies perceived as direct beneficiaries of AI capital spending and electrification, and what the market has come to call “AI winners.” Micron Technology, Inc., one of the three largest memory chip manufacturers globally and a stock we added to the Fund during the quarter, appreciated 242%, vaulting it into the trillion dollar market cap club alongside fellow semiconductor giants NVIDIA Corporation and Broadcom Inc. This market concentration is arguably defensible on fundamentals: the S&P 500 posted one of its strongest quarterly earnings results in five years, driven largely by the Magnificent Seven, which reported blended earnings growth of more than 60%, well above the low double-digit growth recorded by the rest of the Index.

Sector performance reflected the quarter’s one-sided market dynamic. IT outpaced all other sectors by a wide margin, accounting for roughly two-thirds of the S&P 500’s gains and advancing 31.8% for the quarter. The next closest sector, Industrials, rose 14.9%. IT strength was concentrated in AI infrastructure builders within the semiconductors and semiconductor materials & equipment sub-industries.

From a style perspective, following three consecutive months of underperformance to begin 2026, growth stocks staged a meaningful comeback in the second quarter, fueled by the AI-driven rally in April and May. Growth outperformed value by a wide margin across all market cap segments. Despite this recent strength, growth has yet to achieve parity with value on a year-to-date basis.
 

Performance

We manage the Fund with an unwavering focus on powerful secular growth trends that disrupt industries and create sustained, profitable growth opportunities—not short-term geopolitical disruptions or sentiment swings, which whipsawed markets during the first half of this year. Transformative secular trends—such as AI; space exploration and technology; autonomous transportation; robotics; digital commerce, media, and finance; advanced therapeutics and minimally invasive surgery—will shape the future and drive long-term investment returns.

As market sentiment shifted, the second quarter proved to be a favorable backdrop for the Fund with AI winners leading the way and a rotation back into growth stocks. The Fund climbed over 27%, outperforming both the Benchmark and the broader-market S&P 500 Index.

The Fund’s outperformance for the quarter was driven entirely by stock picking—across a diversified set of innovation and secular growth leaders. Our stand-out contributor to absolute and relative performance during the period was launch and satellite broadband groundbreaker Space Exploration Technologies Corp. (SpaceX), which conducted a successful initial public offering during the quarter. SpaceX remains the largest holding in the Fund and the largest overweight position versus the Benchmark. In the IT sector, our systems software investments contributed to absolute and relative performance, with Datadog, Inc. (cloud-based observability and security platform leader), Snowflake Inc. (cloud data platform pioneer), and Rubrik, Inc. (cloud data security and cyber resilience leader) leading the way. Several of our AI-infrastructure and semiconductor investments also contributed to absolute and relative performance, including Taiwan Semiconductor Manufacturing Company Limited (the world’s advanced-semiconductor manufacturing champion), indie Semiconductor, Inc. (automotive semiconductor innovator), Coherent Corp. (optical technology leader), ASML Holding N.V. (the world’s dominant provider of advanced lithography semiconductor manufacturing equipment), Micron, and Broadcom. The Fund also benefited from not owning Microsoft Corporation, which was among the Magnificent Seven and software laggards during the period.

The Fund’s Health Care investments performed well, with stand-out absolute and relative performance from Hinge Health, Inc. (digital musculoskeletal and physical therapy platform trailblazer), as well as a positive contribution from argenx SE (immunology and antibody-based medicine innovator).

Within Industrials, our AI data center power investments also performed well, with exceptional gains and relative performance from Forgent Power Solutions, Inc., a leading manufacturer of custom electrical distribution equipment serving data centers, the power grid, and energy-intensive industrial applications.

While the Fund’s semiconductor investments added value overall, our non-ownership of Advanced Micro Devices, Inc. was the largest drag on relative performance. Among Fund holdings, the top relative performance laggards included Spotify Technology S.A. (global streaming music and content leader), GDS Holdings Limited (leading China data center operator), and Guidewire Software, Inc. (leading cloud-based property and casualty insurance vendor).
 

AI Update

The AI Buildout Continues to Exceed Expectations

The largest companies in the world continued to expand their AI infrastructure investment plans during the second quarter. Over the past three months, Alphabet Inc. raised its 2026 capital expenditure guidance to $180–$190 billion, Microsoft lifted its outlook by $25 billion to roughly $190 billion, Meta Platforms, Inc. raised its range to $125–$145 billion, and Amazon.com, Inc. reaffirmed approximately $200 billion. Together, these hyperscalers are on track to spend more than $700 billion this year, up 74% compared to last year. On top of this, emerging “neoclouds” — specialized AI cloud providers — have continued to grow rapidly, now representing approximately 5% of the overall cloud market, and are investing aggressively as well.

To put this spending in perspective, consider the cost of a single gigawatt of AI compute capacity — the industry’s new unit of account. Estimates from NVIDIA, OpenAI, and Wall Street models range from $35–$60 billion per gigawatt all-in across current chip generations, and NVIDIA’s CEO recently suggested next-generation facilities could approach $80–$100 billion as memory and component shortages inflate costs. Roughly 60–70% of each gigawatt is compute and networking — semiconductors, switches, and optical interconnects — with the balance going to powered shell, electrical distribution, cooling, and backup generation. Each successive chip generation delivers a multiple more intelligence per dollar and per watt, which is why the industry keeps investing despite rising unit costs.

Is all this capacity justified? Two forms of evidence suggest demand is tangible and still running ahead of supply. First, the revenue ramps at the AI labs continue to defy precedent: Anthropic’s annualized revenue run rate reached $47 billion in May, up from $30 billion in early April and $9 billion at year-end 2025, while OpenAI’s enterprise business continues to scale toward parity with its consumer franchise. Second, contracted backlogs at the cloud providers — legally committed future revenue, not forecasts — are growing at extraordinary rates. In the first quarter, Google Cloud’s backlog nearly doubled sequentially to over $460 billion, Amazon Web Services’ backlog grew 93% year-over-year to $364 billion, excluding a new $100 billion OpenAI commitment announced after the period, and the combined backlog across Alphabet, Amazon, and Microsoft now exceeds $1.4 trillion. Every major cloud provider described itself as compute constrained — unable to build capacity fast enough to serve contracted demand.
 

The Agentic and Inference Inflection

The composition of AI compute demand is also shifting. Inference — the everyday use of trained models by businesses and consumers — now accounts for roughly two-thirds of all AI compute, up from approximately one-third in 2023, and for the first time the inference market is growing faster than training. The driver is agentic AI: systems that plan, use tools, verify their own work, and complete multi-step tasks with limited human supervision. Coding remains the flagship use case, and AI agents now write a meaningful share of the world’s new software. But agents are also proliferating across customer service, research, financial analysis, and back-office workflows. Because an agent may make 10 or 20 model calls to complete a single task, agentic workloads consume 5 to 30 times more tokens — the basic unit of AI output — than a simple chatbot query. The result is explosive token demand. Alphabet, for example, disclosed that its AI models are now processing 3.2 quadrillion tokens per month, up more than 300 times from consumption rates two years ago, and Amazon’s Bedrock service processed more tokens in the first quarter than in all prior periods combined.

This inflection is also diversifying the compute landscape. Varied inference workloads — short chats, long-context coding sessions, always-on monitoring agents — favor different silicon architectures, which is why custom application-specific integrated circuits, inference-optimized chips, and disaggregated designs are proliferating alongside NVIDIA’s platforms. Meanwhile, competition between frontier labs and rapidly improving open-source models has driven the price per token down as much as 60% to 80% on flagship models over the past 18 months — and falling prices have accelerated usage rather than dampened spending. Whatever the quarter-to-quarter noise, token demand is heading up and to the right.
 

Measuring the Returns on AI

We have long argued that the most important question in AI is not the race to artificial general intelligence but the utility and value it delivers. The evidence of measurable economic return keeps building. In a recent report, Morgan Stanley analyzed more than 17,000 earnings call and conference transcripts and found that 40% of companies its analysts identify as “AI Adopters” cited at least one quantifiable benefit from AI in the second quarter — nearly double the 21% reported a year ago — while roughly 25% of the broader S&P 500 Index did, up from 14%. The benefits are concentrated in financial impact and productivity: HP is targeting $1 billion in annual run-rate savings from AI-enabled automation, Verizon has captured over $200 million in energy savings by deploying AI across its network, and Equifax has embedded $75 million of AI-driven cost reductions directly into its guidance. Morgan Stanley’s companion survey of 935 corporate executives found companies realized approximately an 11% net productivity improvement over the past 12 months. Against the firm’s estimate that full AI adoption could ultimately generate approximately $920 billion of annual economic value for S&P 500 companies — roughly 28% of expected 2026 pretax earnings, largely through operating expense efficiency and productivity gains — we believe the returns on AI are early but increasingly demonstrable, and that companies with pricing power and proprietary data should capture a disproportionate share.
 

The AI IPO Era Begins

The public markets opened their doors to the AI era in dramatic fashion during the quarter. On June 12, SpaceX — the Fund’s largest holding — completed the largest IPO in history, raising $75 billion at a $1.77 trillion valuation before rallying to a market capitalization above $2 trillion. The company’s prospectus cited a combined addressable market of $28.5 trillion spanning launch, broadband, and AI, and disclosed that the majority of its first quarter capital expenditures were directed at AI. Anthropic, days after closing a $65 billion funding round at a $965 billion valuation, confidentially filed for an IPO on June 1; OpenAI followed shortly thereafter, with both expected to list within the next several quarters. The common thread across these landmark offerings: extraordinary revenue growth, enormous addressable markets, and a need for public capital to fund the AI buildout. We expect these listings to broaden the investable AI universe and replace speculation with evidence — which suits our research-driven approach just fine.

Below is a partial list of the secular megatrends we focus on:

• AI
• Semiconductors
• Cloud computing
• Software-as-a-service
• Digital media/entertainment
• Targeted digital advertising
• E-commerce
• Targeted medicine/therapies
• Minimally invasive surgical procedures
• Cybersecurity
• Electric vehicles/autonomous driving
• Electronic payments
• Robotics
• Space technology

We continue to run a high-conviction portfolio with an emphasis on the secular trends cited and listed. Among others, during the second quarter we initiated or added to the following positions:

• Semiconductors and cloud compute services: Micron Technology, Inc. and Cerebras Systems Inc.
• Power generation and energy infrastructure/aerospace: Fervo Energy Company and DPC Holdings Limited
• Digital content, commerce, and AI services: Alphabet Inc., Spotify Technology S.A., and Shopify Inc.
• Health care real estate and digital services: Welltower Inc.
• Cybersecurity and systems software: Rubrik, Inc., CrowdStrike Holdings, Inc., and Cloudflare, Inc.

Space Exploration Technologies Corp. was the Fund’s top contributor in the quarter, following its record-breaking initial public offering—the largest in history. We first invested in SpaceX in 2020, when it was still viewed primarily as a daring launch pioneer. That year, the company completed just 26 Falcon 9 launches, with a maximum of 7 re-flights on any single booster. Starlink was in its earliest days, with fewer than 1,000 satellites in orbit and only a few thousand beta testers. The Dragon crew capsule had flown two astronauts on its historic demonstration mission, and there was no Starlink Mobile, no Starship flights, and no AI business to speak of. Fast forward to 2026, and SpaceX has become an unparalleled technology and infrastructure powerhouse. In 2025 alone, it executed over 170 launches, including 165 Falcon 9 missions, with individual boosters now achieving up to 35 re-flights. The Starlink constellation has grown to more than 10,000 satellites — representing over 75% of all active maneuverable satellites in orbit — and now serves more than 12 million broadband subscribers globally. The company launched Starlink Mobile with over 650 dedicated satellites and has partnered with telecommunications operators to connect billions of people and devices worldwide. Dragon has now safely transported nearly 80 people to space.

Beyond space infrastructure, SpaceX has made a bold and strategic move into AI. Through its integration of xAI, SpaceX deployed the world’s largest coherent compute cluster faster than any competitor and has monetized a significant portion of that capacity through agreements with partners including Anthropic and Alphabet, generating over $26 billion in annualized revenue. The acquisition of AI coding pioneer Cursor further strengthens its AI competitive position by expanding talent, proprietary training data, and presence in the large and rapidly growing agentic coding market. Looking ahead, Starship’s increasing operational maturity will unlock dramatic growth in launch cadence and payload capacity, enabling the rapid deployment of next-generation V3 Starlink satellites to expand coverage and throughput. More importantly, Starship will power the deployment of orbital AI compute infrastructure at gigawatt scale—taking advantage of continuous solar power, efficient radiative cooling, and freedom from terrestrial constraints. Combined with vertical integration into chip design and manufacturing through the Terafab initiative, SpaceX is positioned for unmatched cost leadership in AI, one of the largest addressable markets in history. Additional long-term opportunities include building a lunar economy through cargo transport and in-space manufacturing, point-to-point Earth travel, space tourism, and laying the foundation for eventual Mars colonization. SpaceX remains the Fund’s largest holding and largest overweight position versus the Benchmark, and we believe it is one of the most extraordinary companies ever built.

Shares of NVIDIA Corporation contributed to performance during the quarter. NVIDIA is a semiconductor and systems company specializing in compute and networking platforms for accelerated computing. The company’s dominant position in AI infrastructure—spanning AI accelerator chips (graphics processing units, or GPUs), systems, software, and high-performance networking solutions—continues to underpin strong fundamentals and robust financial results. The company delivered another strong earnings report and raised its forward outlook meaningfully above prior expectations. Demand for NVIDIA’s products remains exceptional—as CEO Jensen Huang remarked on the first quarter earnings call, demand has gone parabolic. In the company's April quarter, data center revenue grew 92% year-over-year to $75.2 billion, with total revenue of $81.6 billion, up 85%; for its July quarter, the company guided for total revenue to grow at least another $9.5 billion sequentially. The company’s offerings retain strong differentiation as the AI compute paradigm evolves from model training to inference to the emerging agentic era, in which AI agents autonomously plan and execute complex tasks. This was particularly evident in the company’s disclosure that it is poised to become one of the world’s largest CPU (central processing units) providers — a product category that has historically been outside its core focus. The company’s next-generation platform, Vera Rubin, remains on track for second-half delivery despite earlier rumors of delays. We retain conviction in NVIDIA as a core long-term holding, viewing its full-stack platform, software moat, and visibility into sustained hyperscaler capital spending as durable advantages supporting a multiyear growth trajectory.

Broadcom Inc. is a global designer and supplier of semiconductor and infrastructure software solutions at the core of modern computing and networking. The company is a leader in high-performance digital and mixed-signal technologies spanning networking, connectivity, storage, and custom silicon accelerators. Through its acquisition of VMware, Broadcom also owns critical software layers used to virtualize and manage large-scale compute environments. Shares contributed to performance during the quarter as the company’s key customer, Alphabet (Google), signaled higher and more durable capital expenditures over the coming years than previously anticipated. Broadcom’s multi-year agreement with Google, extending through 2031, validates its entrenched position within the Google silicon ecosystem and establishes it as a primary beneficiary of Google’s AI-related infrastructure spending. Beyond Google, several other customers showed incrementally positive signs in their custom silicon adoption journeys. Anthropic, following its recent commercial success, is planning for significantly larger compute requirements, which should translate into a growing custom silicon base over the next several years. OpenAI, working with Broadcom, taped out its first inference chip—codenamed Jalapeno—in a record nine months, and appears set to build 10 gigawatts of AI infrastructure using custom silicon this decade. Meta continues to hold a constructive view on AI investment and the associated infrastructure buildout. Additionally, Apple signed a multi-year agreement with Broadcom spanning multiple product lines—encompassing not only conventional radio frequency components and next-generation wireless connectivity technologies, but also custom silicon across multiple generations of Apple products. As CEO Hock Tan has noted, custom silicon accelerators are on pace to match GPU units in volume by next year, and Broadcom, as the category leader, is well positioned to be the primary beneficiary of this transition. We maintain our conviction in Broadcom. The company is uniquely positioned to capture a dominant share of the custom silicon market and faces no credible terminal risk to its custom silicon franchise over the foreseeable future. We continue to hold the stock and believe Broadcom is on a path to becoming one of the largest technology companies in the world.

Despite solid operating results, shares of GDS Holdings Limited declined during the second quarter. Weighing on sentiment were management’s communications regarding full-year guidance, a material step-up in capital expenditures over the next several years, and a slight delay in the expected timing of the company’s growth inflection. GDS develops and leases data center space to leading global technology companies—including Alibaba, Tencent, ByteDance, Microsoft, Google, and Oracle—under long-term arrangements. We recently hosted CEO and founder William Huang and CFO Daniel Newman at our offices and continue to believe the best days for the company lie ahead, supported by powerful secular tailwinds: the early stages of cloud adoption in Asia, continued data growth, rising AI demand, and global power constraints that are sustaining pricing power. Rather than focusing on the precise quarterly timing of the growth inflection, we remain focused on the long term and see increasing evidence of a building AI wave in China through significant bookings growth. We also see material and underappreciated value in GDS’s stake in its spun-out international subsidiary, DayOne, which we expect to list publicly at a significantly higher valuation than its most recent private capital raise, supported by continued bookings and cash flow momentum and highly visible take-or-pay revenue ramp timelines.

Shares of property and casualty insurance software vendor Guidewire Software, Inc. declined during the quarter as a small number of deals slipped from its fiscal third quarter into its fiscal fourth quarter, coinciding with broader fears of AI-driven disruption that pressured valuation multiples across the application software sub-industry. We believe the deal slippage was purely a timing issue and that all affected deals have since closed. Guidewire’s InsuranceSuite platform serves as the core system of record for insurance carriers—the single source of truth for policies written, claims processed, premiums collected, and payments made. The complexity of insurance policies, their highly regulated nature, and the fact that they exist exclusively within Guidewire rather than in physical form makes this system of record particularly critical and therefore highly valuable. We believe the core system opportunity alone represents nearly $20 billion of annual recurring revenue, or approximately 20 times Guidewire’s current scale. We also believe AI will be a significant tailwind for the company, meaningfully expanding this opportunity by enabling automation and intelligence layered on top of the core system of record. We are already seeing Guidewire bring new AI-enabled capabilities to market and sign customers, and we expect adoption to accelerate over the coming year. Finally, we expect Guidewire to benefit from the same internal productivity enhancements that AI is delivering across industries, helping the company support faster growth at lower cost and, ultimately, better profitability.

We elected to exit our position in real estate information and marketing services provider CoStar Group, Inc. after a successful multi-decade investment. Performance has been challenged over the past several years as the company has invested aggressively to build out its Homes.com residential real estate marketing platform. Revenue traction for Homes.com has been slow to materialize, and we are concerned that recent shifts in market structure will make it incrementally more difficult for CoStar to grow a residential business commensurate with its level of investment. Shares have also been pressured by the sector-wide AI disruption concerns weighing on the broader software and information services ecosystem.

We invest in secular growth and innovative businesses across all market capitalizations, with the bulk of the portfolio landing in the large-cap zone. Morningstar categorizes the Fund as U.S. Large Growth. As of the end of the second quarter, the largest market cap holding in the Fund was $4.8 trillion and the smallest was $500 million. The median market cap of the Fund was $90.4 billion, and the weighted-average market cap was $1.7 trillion.

To end the quarter, the Fund had $2.1 billion of assets under management. We had investments in 46 unique companies. The Fund’s top 10 positions accounted for 63.1% of net assets.

Micron Technology, Inc. is an industry leader in memory and storage solutions that are increasingly central to the advancement of AI. The core investment thesis begins with a structural bottleneck: AI inference applications are architecturally constrained by memory bandwidth, and the scaling laws that have historically governed compute performance have expanded to encompass memory capacity as well—meaning material gains in AI performance can be captured by deploying larger, faster memory. This has created what we believe is a durable, multi-year demand tailwind that will continue to outstrip supply, supported by the incremental returns on memory investment remaining exceptionally high. Near-term supply constraints and the resulting price increases are constructive for Micron’s financials, but what interests us more is the structural evolution of the business itself. Memory players, and Micron specifically, are entering into long-term strategic supply agreements with customers that lock in visible demand and meaningfully dampen the cyclicality that has historically characterized the industry. Meanwhile, each successive generation of memory technology moves Micron further from commodity territory: while high-bandwidth memory co-packaged with AI accelerators has already added stickiness and differentiation, the next generation of solutions will embed controller logic within the memory itself—an architectural shift that makes memory a far more proprietary and defensible product than in prior cycles. We believe Micron is a fundamentally different business than it was even a few years ago: one positioned not only to grow earnings significantly, but to command a higher valuation multiple as the market comes to appreciate the durability and strategic importance of its competitive position.

During the quarter, we initiated a position in Fervo Energy Company, a pioneer in enhanced geothermal systems (EGS). While geothermal energy has many desirable characteristics—including baseload power with minimal operating costs and no emissions—the industry has historically been constrained by limited geographic applicability, requiring proximity to areas of volcanic activity, and by the challenges and costs of resource exploration, including the risk of unproductive “dry wells.” EGS addresses these limitations by leveraging techniques from innovations in oil and gas drilling — horizontal drilling and hydraulic fracturing — to unlock additional resources and drive down development costs. Fervo is by far the market leader in this segment of the geothermal industry, having developed proprietary drilling techniques and plant designs over the eight years since its founding, and the company is on the cusp of bringing its first commercial plant online later this year. Fervo already has 658 megawatts under power purchase agreements with both utilities and hyperscalers and believes it has access to at least 42 gigawatts of available resource across hundreds of thousands of land leases it acquired well before competitors were paying attention to the EGS opportunity. The company also has a clear path to drive costs per well down to $3,000 per kilowatt over time—having already reduced costs by thousands of dollars per kilowatt from its initial wells—which would put its levelized cost of electricity production at or below competing alternatives. Given the highly desirable attributes of geothermal power, we believe that as Fervo proves out its first commercial plant, it will see accelerating contracting and deployment activity over the next decade, creating meaningful upside potential for shareholders.

During the quarter, we added to our position in Alphabet Inc., the parent company of Google—the world’s largest search and digital advertising franchise—as well as YouTube and Google Cloud Platform, one of the three leading hyperscale cloud infrastructure businesses globally. Alphabet also owns DeepMind, one of the premier AI research organizations in the world. Alphabet generates approximately $174 billion in trailing operating cash flow, a funding advantage that becomes increasingly important as the AI race enters a phase of high capital intensity. Our conviction deepened this quarter as several developments reinforced our view that Alphabet is one of the most strategically advantaged businesses in the AI era. At Google I/O 2026, management disclosed that Alphabet now processes 3.2 quadrillion tokens monthly—seven times increase year over year—with the Gemini app surpassing 900 million monthly active users, more than doubling in a year. AI is proving additive rather than cannibalistic to core search. Search revenue grew 19% year-over-year in the first quarter, and CEO Sundar Pichai noted that AI features are driving more search usage, not less. Google Cloud accelerated to 63% year-over-year revenue growth—a five-year high—with operating margins expanding to 33% and the cloud backlog nearly doubling sequentially to approximately $462 billion, anchored in part by a landmark multi-year tensor processing unit (TPU)⁵ commitment from Anthropic, affirming that Alphabet’s custom silicon is winning the confidence of the most demanding AI infrastructure customers. What distinguishes Alphabet from every other competitor in the AI landscape is the breadth of its vertical integration: custom eighth-generation TPUs optimized for both training and inference, frontier Gemini models, the Antigravity enterprise agentic coding platform, and consumer distribution across 13 products with more than 1 billion monthly active users each—including five with more than 3 billion. No pure-play AI lab or cloud competitor replicates this stack, and we believe the cost and distribution advantages it confers widen over time. We see a long runway for growth as Google Cloud accelerates, Gemini monetization deepens across search and subscriptions, and this full-stack advantage compounds—a combination we believe bodes well for long-term shareholders.

During the quarter, we participated in the IPO of Cerebras Systems Inc., a semiconductor design and compute services company building AI accelerators around one of the most differentiated architectures in the market. We developed a strong familiarity with the Cerebras management team during its final two years as a private company, including multiple meetings and a visit to its headquarters just months before the offering. As AI inference has become the dominant and fastest-growing component of AI compute demand, a critical sub-segment has emerged: fast inference, or the speed at which an AI model generates responses, measured in tokens per second. As developers build applications and agents that chain multiple models together, demand for tokens per second is growing by orders of magnitude—and Cerebras has designed a specialized chip exceptionally well suited to serve this need. Unlike most semiconductor startups, Cerebras holds a genuine performance advantage for fast inference workloads, and customer engagements speak louder than benchmarks: a multi-year agreement with OpenAI for 750 megawatts of inference capacity valued at more than $20 billion—with an option to add another 1.25 gigawatts—reflects real confidence in both the fast inference market and Cerebras’ differentiated value proposition, and a new partnership to bring fast inference to Amazon Web Services further validates the opportunity. The company is currently pursuing both direct AI systems sales and cloud services to drive adoption. We expect the AI infrastructure buildout to sustain its momentum, with a large and growing share of workloads requiring fast inference, and as the category leader, Cerebras is well positioned to benefit while continuing to innovate to hold its lead. We are clear-eyed about the risks—the company must execute on the OpenAI deployments, develop architectural solutions as workloads evolve and potentially demand significantly more memory, and depend on customers continuing to pay a premium for faster inference—but on balance, we find the setup compelling and initiated a position during the quarter.

During the quarter, we added to our position in Welltower Inc., which owns and operates senior housing communities in the U.S. and internationally. While Welltower screens as a real estate business, we view it as the intersection of hardware, real estate, and software—its proprietary operating platform and data analytics capabilities. Rolling out this software layer creates meaningful structural upside to both operating margins and occupancy through enhanced asset management, proprietary analytics, and new initiatives such as amenity-based pricing. We recently hosted the entire Welltower executive team in our offices and came away more encouraged by the multi-dimensional growth opportunity ahead—particularly the early monetization of its proprietary data analytics platform and the continued rollout of the Welltower Business System. The company has deliberately recruited senior talent from both technology and real estate to drive this transformation, and we believe CEO Shankh Mitra and his management team are disciplined capital allocators focused on driving accretive value per share. The broader industry backdrop is among the most favorable in years: demand is supported by powerful demographic tailwinds, with the 80-plus population growing at a 4% to 5% compound annual rate over the next five years, well above the 2% rate that followed the global financial crisis, while supply remains structurally constrained by declining construction starts, unattractive developer economics, and a five-plus year entitlement and build timeline. The constrained financing environment for senior housing should continue to generate an active external growth pipeline at an attractive basis. Putting it all together, we see a path for earnings to more than double over the next five years, creating attractive long-term return prospects for the Fund.

We exited our position in ServiceNow, Inc., which was a successful long-term investment for the Fund, and redeployed the capital into the software and digital services names discussed in the Review and Outlook section above.

We trimmed our Datadog, Inc. position to a lower portfolio weight after the stock rose 121% in the second quarter. We retain conviction in Datadog’s long-term growth opportunity, durable competitive advantages, and innovative and disciplined management team.

We sold out of our long-term investment in CoStar Group, Inc., as described above.

I remain confident in and committed to the strategy of the Fund: durable growth based on powerful, long-term, innovation-driven secular growth trends. We continue to believe that noncyclical, durable, and resilient growth should be part of investors’ portfolios and that our strategy will deliver solid long-term returns for our shareholders.

Sincerely,