Baron Fifth Avenue Growth Fund | Q3 2026

We had another great quarter.

Baron Fifth Avenue Growth Fund (the Fund) gained 6.6% (Institutional Shares) during the third quarter, outperforming the 0.9% gain of the Russell 1000 Growth Index (R1KG) and the 2.3% gain of the S&P 500 Index (SPX), the Fund’s benchmarks.

Year to date, the Fund is up 19.0% compared to gains of 6.3% and 12.8% for the Fund’s benchmarks, respectively.

Despite the heightened volatility, the markets continued to grind higher, eking out a modest gain in the quarter. The Fund “behaved” particularly well during this time, accomplishing a rare feat of outperforming both on the way down and on the way up, with strong absolute and relative returns. We booked a solid 6.6% gain, generating 5.6% excess return over the Russell 1000 Growth Index, the Fund’s primary benchmark.

The third quarter opened with a sell-off in AI buildout stocks, triggered by a confluence of rumors, news, and events. The Situational Awareness hedge fund that held large, leveraged bets on semiconductors and AI infrastructure stocks had collapsed. Unconfirmed rumors that Meta was considering renting out some of its AI compute caused the talking heads to speculate that Meta’s internal AI demand had weakened. Data showing that open-source models were gaining token share from the frontier companies’ proprietary LLMs was accompanied by token price declines. Widely circulated speculation that Anthropic and OpenAI’s growth was decelerating culminated in a leak that Anthropic had reached $65 billion of annual recurring revenue (ARR) in July⁵, adding “only” $18 billion in two months – roughly a Texas Instruments’ worth of revenues. Obviously disappointing, eye-roll emoji.

By the end of July, semiconductors, which had become a chunky 28% of the Index, had fallen 7.0%, semiconductor materials & equipment stocks dropped 32.8%, and electronic components securities declined 27.0%. Some of the biggest darlings of the first six months such as Micron, Sandisk, Corning, and Lam Research declined anywhere from the high 20% to mid-40% range. AI-buildout related Industrials stocks also saw a sharp sell-off, turning the biggest winners from the first half into major laggards. GE Vernova, Bloom Energy, and Caterpillar saw their stocks decline from the mid-teens to low-30% range. Fast money investors also sold SpaceX, which fell 36.6% in July after a successful IPO in June.

Despite a large position in SpaceX (our second largest active weight), the Fund declined just 3.9% in July, outperforming the R1KG’s 4.8% drop. This was largely due to the continued rebound in our software holdings Cloudflare, Samsara, and Snowflake, each of which posted double-digit gains in July as the early 2026’s SaaSPocalypse faded into a distant memory. Amazon also rallied sharply after posting great quarterly results with accelerating revenue growth that reached 20% year-on-year, driven by 37% growth in AWS, it's fastest in 18 quarters (when it was half the size), with operating income growth of 43% and expanding margins.

The AI-buildout sell-off abruptly ended in August as strong earnings from hyperscalers, neoclouds, and semiconductor companies did not disappoint. Hyperscalers reported accelerating growth at massive scale with their cloud revenue run-rate rising 50% year-on-year to $420 billion⁶. Neoclouds reported signing deals at rising (rather than declining) prices, with shorter payback periods and higher ROI. CoreWeave (which we do NOT own) secured a contract to rent its six-year-old Ampere GPUs through 2029, while Nebius (which we own in our Global Opportunity Fund) disclosed price increases for both old and new GPUs as demand continued to strengthen. This momentum continued in September. NVIDIA reported $96 billion in revenue, with growth accelerating to 106% year-on-year, a broader customer base, and growing content per GW. The company also guided to 70% growth for next fiscal year, significantly above expectations noting that supply constraints prevented an even higher outlook! Broadcom likewise reported strong results with 86% year-on-year revenue growth, and projected that its AI business would double next fiscal year and then double again the year after! We have not seen this kind of growth… profitable growth… at scale… before. Ever!

The Fund also outperformed in August, gaining 6.4%, compared to the 3.7% gain for the R1KG. SpaceX led the recovery, rising 32.6% after strong quarterly results, with revenues that nearly doubled year-on-year, adjusted EBITDA that nearly tripled, accelerating the pace of its AI buildout as ROI continues to improve. September proved to be a strong month as well, with the Fund returning 4.3%, outperforming the R1KG’s 2.2% gain.

For the third quarter overall, stock selection accounted for 476bps of the Fund’s 561bps of outperformance, while sector allocation contributed 86bps. Information Technology (IT) was our strongest sector, adding 277bps entirely through stock selection. The leading sub-industries were systems software, led by CrowdStrike, Cloudflare, and Snowflake; semiconductor materials & equipment, where our holding ASML fell just 8.8% versus the R1KG sub-industry’s 28.2% decline; and internet services & infrastructure, supported by Shopify’s 29.9% gain. Together, these three sub-industries contributed 326bps of outperformance. Stock selection was also strong in Health Care, where Illumina rose 55.8%, and Consumer Discretionary, where Amazon and MercadoLibre gained while the Index sector fell 11.1%. Not owning investments in Industrials, Consumer Staples, Utilities, or Real Estate, all of which declined in the quarter, added another 178bps to relative returns. Communication Services partially offset these gains, costing 21bps as Meta’s 28.8% rise was not enough to offset the 11.7% decline for SpaceX.

From an absolute, stock-specific perspective, the Fund had 16 contributors against 11 detractors. Similarly to last quarter we had many big winners with 9 investments generating over 50bps each, including NVIDIA, Cloudflare, Meta, CrowdStrike, Shopify, and Illumina, which contributed over 100bps each. SpaceX and Tesla were the only meaningful detractors, costing us 210bps combined.

During the quarter, we traveled to the Bay Area to meet with public and private software and semiconductor companies, VC and PE firms, and a variety of industry practitioners that are involved in the AI buildout across energy, construction, and financing. These trips are always a good source of data from a wider group of industry participants. The main purpose is to seek disconfirming evidence, test our own assumptions and hypotheses, and to broaden and sharpen our industry perspectives. It is clear that at this early stage of the AI S-curve what we do not know far outweighs what we do. Nevertheless, we came away with several important observations:

• Demand for AI tokens remains robust and is accelerating. Strong token demand was the most consistent theme in our conversations. Companies are seeing Jevons paradox in action – as the price per intelligence task declines, AI adoption accelerates. According to research by epoch.ai⁷, the cost for a given level of intelligence has fallen by a factor of 13 each year since 2023 – a rate 4x faster than DNA sequencing, 6x faster than compute, 18x faster than lithium batteries, and 54x faster than electricity! Companies of all sizes described sharply increasing token consumption as the price per token declined. These accounts align with the rapid growth reported by hyperscalers and the doubling of token consumption volumes now, approximately every eight weeks⁸.
• Near-term data center construction delays are a growing risk, but the growing deployment costs and complexity help guard against an overbuild as a self-correcting mechanism. The unprecedented scale of the buildout requires coordination across permitting and local politics, power equipment, utilities, construction, skilled labor, capital, servers, memory, networking, and more. The range of outcomes on the amount of GWs brought online over the next several years remains very wide. Our discussions reinforced the value of companies that can execute real-world, large-scale buildout projects, such as SpaceX, and those with substantial installed AI capacity, including Meta, Amazon, and increasingly SpaceX. If supply grows slower than expected, the widening gap between demand and supply will increase the value of each available GW, while reducing the risk of an overbuild. Slower terrestrial capacity growth would also make SpaceX’s development of orbital compute more important. This reinforced a part of our theses that we articulated in the last quarterly letter:

“Over time, as building data centers on Earth becomes more challenging and expensive because of regulation, power and cooling constraints, land scarcity, and construction costs, orbital data centers could benefit from effectively unlimited power, unlimited space, economies of scale, and a simpler regulatory framework.”

• Open weight models’ adoption is growing. As their capabilities meet more use cases, companies are adopting multiple models rather than standardizing on one. Many are post-training open models on proprietary data to protect their IP and improve efficiency for their specific/custom needs. We continue to expect both frontier and open-weight models to serve distinct use cases. In our view, the adoption of open-weight models will likely benefit AI infrastructure and its picks-and-shovels providers as these models still need AI servers to generate tokens, while avoiding frontier providers’ higher margins lowers the cost per intelligence task and further supports Jevons paradox.
• AI-native startups are growing at an unprecedented pace. While many investors are familiar with the extraordinary growth rates of Anthropic and OpenAI, we came away with greater appreciation of the diffusion of AI, even at this early stage. AI unicorns including Sierra, ElevenLabs, Cognition, Legora, Harvey, Lovable, and Vercel are growing at triple-digit rates on revenue bases of hundreds of millions of dollars. These companies are reaching $100 million, $200 million, $400 million, and $1 billion revenue milestones faster than any SaaS companies we have seen before.

For much of the past three years, investor sentiment has swung between two extremes: “AI is a bubble with no ROI” and “AI is so powerful that software is dead and labor will be disrupted.” These rapid shifts, compounded by the early stage of the AI S-curve, have fueled significant volatility, with correlations across entire stock groups approaching 1, at either extreme. For bottom-up fundamental investors who operate as long-term owners of businesses this has created real and substantive opportunities: buying systems software and cybersecurity companies in late 2025 and early 2026, buying AI stocks during the early-2025 DeepSeek scare, or holding Meta when the prevailing narrative declared it dead. This is how our process, when executed well, can add value and generate alpha during the periods of extreme volatility and stress.

NVIDIA Corporation is a global leader in accelerated computing designing the semiconductors, software, networking, and systems used to build and deploy AI. Shares rose 14.3% during the quarter, gaining ground even amid a broader sell-off in semiconductor stocks. NVIDIA reported another excellent quarter with revenue growth accelerating to over 100% year-on-year at a massive scale of $96 billion, with the company guiding to 70% revenue growth for next fiscal year, far ahead of analyst expectations. Notably, management characterized this outlook as supply-constrained, with customers’ own forecasts pointing to a growth rate of approximately 100% year-on-year. Demand has also broadened, with customers beyond the largest hyperscalers, from sovereign AI and neoclouds to enterprises, now accounting for nearly half of data center revenue and growing faster than hyperscalers (though hyperscalers also continue to grow rapidly). In addition, each new platform captures more revenue per gigawatt as NVIDIA sells central processing units, networking, and inference systems alongside its graphics processing units and as each generation provides a higher throughput of tokens per MW of power. We retain conviction in NVIDIA as a core long-term holding. We continue to view its rapid innovation velocity, full-stack platform, extreme co-design capabilities, software, eco-system and supply chain management, as durable advantages that support continued growth in earnings power over time as AI adoption diffuses across the economy.

Shares of Cloudflare, Inc., which operates one of the world's largest networks, delivering security, performance, and developer solutions, rose 41.5% during the quarter as investors increasingly view the company as foundational infrastructure for the agentic internet. In the second quarter, more than half of the traffic across Cloudflare's network was non-human for the first time, and the company is building the controls, developer tools, and payment rails to secure and monetize that shift. A broader cybersecurity recovery added a tailwind, as new AI-enabled threats heightened urgency among security buyers with partners reporting a broad-based acceleration across Cloudflare's portfolio. The company also reported strong growth in fundamentals: revenue growth accelerated to 36% year-on-year, large customers grew 27% with record net additions, and nearly two million developers joined its Workers platform in the quarter alone. We believe Cloudflare's scale and network architecture give it a durable advantage as AI agents reshape how the internet is being used. We remain confident in the long-term opportunity while remaining mindful of valuation following the stock’s substantial appreciation.

Shares of Meta Platforms, Inc., the owner of Facebook, Instagram, WhatsApp, and Threads, contributed during the quarter, rising 28.8%. Second quarter results showed the core ad engine is getting stronger: U.S. advertising revenue growth accelerated to 31% as new AI ranking and recommendation models lifted conversions, and Advantage+, its AI-driven ad suite, passed $75 billion in annual revenue run-rate. The bigger catalyst came in September with the launch of Muse, Meta's personal AI agent, which reached No. 1 on the U.S. App Store within days. At its conference, Meta Connect, the company introduced real-time voice, a Mac app with computer use, and commerce partners including Stripe, Shopify, PayPal, Walmart and Best Buy, so that Muse can complete purchases. This further strengthened Meta’s positioning and added new potential avenues for monetization, while also improving its contextual understanding of users, which should further advance Meta’s ability to match the right ad to the right user at the right time. Meta also brought Muse to its smart glasses and to Charm, a new handheld device. We believe current earnings understate Meta's earnings power as it invests ahead of revenue, and we retain strong conviction that its data, compute, distribution and founder leadership will make it a leader in consumer AI. We also believe AI will remain a tailwind for Meta’s core business, helping the company sustain a long duration of growth.

Space Exploration Technologies Corp. (SpaceX) develops and launches advanced rockets, satellites, and spacecraft, with the long-term goal of making humanity multi-planetary. The stock detracted from performance as shares pulled back 11.7%, despite strong underlying fundamentals. We remain long-term shareholders and believe the company is in the early stages of a multi-decade growth opportunity. SpaceX’s first public-company results reinforced our conviction, with revenue nearly doubling year-over-year and losses narrowing sharply on high incremental margins. Growth was driven by rapidly scaling, multibillion-dollar compute-hosting deals with AI leaders, including Anthropic and Google, alongside Starlink momentum. Management guided to a robust $100 billion annualized revenue run rate by year-end 2026 and forecasted up to five times more AI compute capacity exiting 2027, well above Street expectations. SpaceX also closed its acquisition of Cursor, a premier enterprise AI-powered coding platform, accelerated product development with Grok Bot and more advanced Grok models, and progressed Starship toward full reusability.

Tesla, Inc. manufactures electric vehicles, solar products, and energy storage solutions and develops real-world AI technologies. Shares fell 15.6% as record second quarter revenue and deliveries were overshadowed by near-term profit pressure from increased investment in key opportunities across autonomy, robotics, and AI. While Tesla is funding this ambitious multi-year investment cycle, investors are anxious to see faster robotaxi scaling. We see momentum across the business, with improving automotive demand and accelerating Full Self-Driving (FSD) adoption. The active base reached a record 1.48 million, up 56% year-over-year, with 55% of North American deliveries now including FSD. Unsupervised robotaxi miles are compounding rapidly, with an impeccable safety record across more than one million miles. Cybercab production is ramping, and given the importance of scaling robotaxis safely, we believe Tesla is right to take a deliberate approach to execution. The next-generation AI5 inference chip will also play a central role in the Optimus program. Together, these developments reinforce the robust long-term physical AI opportunity ahead, supporting our conviction to remain shareholders.

Taiwan Semiconductor Manufacturing Company Limited (TSMC) is the world's largest contract chipmaker and the leading manufacturer of advanced logic semiconductors used in modern AI accelerators. Shares corrected 4.3% during the quarter as a result of the broader correction in AI buildout stocks and as investors took profits after the stock’s strong first half performance. Additionally, investor focus shifted to near-term headwinds in the shape of increasing capex and a potential margin dilution from the steep ramp of TSMC’s 2-nanometer node. Despite these concerns, underlying business performance remained strong. Second quarter net income rose nearly 80% year-over-year on a very strong gross margin, and management raised its full-year outlook to dollar revenue growth of slightly above 40%. We retain conviction in TSMC as a core long-term holding and view its near-monopoly positioning in leading-edge chip manufacturing, pricing power, deep and diverse customer relationships, ability to innovate rapidly as chip complexity continues to rise, and its economies of scale as durable competitive advantages that support a long duration of growth.

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

As of September 30, 2026, the top 10 holdings represented 66.3% of the Fund’s net assets, and the top 20 represented 92.8%. We exited the quarter with 26 investments, one less than we had at the end of last quarter.

As of the end of the quarter, IT and Communication Services represented 75.2% of the Fund’s net assets. Consumer Discretionary, Health Care, and Financials accounted for 24.1%, while the remaining 0.7% was held in cash.

During the third quarter, we continued to build our newer positions in the electrical component supplier, Amphenol, the pharmaceutical giant, Eli Lilly, and the fast-inference AI infrastructure provider, Cerebras.

To fund these purchases, we sold our stub position in the Korean e-commerce platform, Coupang. We also trimmed several existing software holdings, Cloudflare, Datadog, and CrowdStrike, as their recent strong stock performance partially pulled forward future returns. For the same reason, we also slightly reduced our position in the leading DNA sequencing platform, Illumina.

We continued building our position in Amphenol Corporation during the third quarter after initiating it earlier in 2026. As a reminder, Amphenol is a leading provider of mission-critical interconnect, sensor, and antenna solutions across a diverse set of end markets from IT Datacom to Industrial, Communications, Automotive, Defense, Commercial Aerospace, and Mobile Devices. While stock price volatility remains elevated driven both by the broader investor sentiment around the AI buildout and by the more specific sentiment around the future value of copper versus optical architectures for AI, the company continues executing at a high level. Second quarter revenues of $8.76 billion were up 55% year-on-year (30% organic growth), which was over $0.5 billion above its guidance. Growth was led by IT Datacom, which was up 89% (63% organic), but all other end markets were up 10% or more organically, showing that Amphenol is not just an AI story. Margins were also exceptionally strong with adjusted EBIT margin of 29.8% (+415bps year-on-year). The company also reported a very strong book-to-bill ratio of over 1.2 times and provided an outlook for the next quarter that was well ahead of expectations, calling for 50% to 52% year-on-year revenue growth. The company also noted that CommScope’s AI-related business has doubled under Amphenol’s ownership already (and with significant synergies still ahead), which reduces the risk around the copper versus optical debate. We continue to believe the market underestimates Amphenol’s long-standing ability to innovate and adapt across cycles and end markets and have therefore continued scaling our position.

We also continued to add to Eli Lilly and Company in the third quarter. Lilly is best known for its GIP/GLP-1 medications for diabetes and obesity. Although awareness and adoption of these medications have grown rapidly in the past few years, the market remains vastly underpenetrated. In the U.S. alone, based on Baron Capital internal research analysis, we estimate that there are roughly 135 million obese or diabetic patients who would qualify for GLP-1 drugs, yet less than 15% are on a GLP-1 drug today. Globally, over one billion people could benefit from GLP-1 therapy, and international markets are even less penetrated. We think these drugs will continue to launch well as awareness and insurance coverage grow, and as Lilly generates more data to show the dramatic impact GLP-1 drugs can have on other medical comorbidities. We think that these drugs will become the standard of care and a $180 billion-plus category. We also view Lilly as a clear leader in GLP-1s as evidenced by continued market share gains, which have now reached more than 60% in the U.S. and nearly 55% internationally. Lilly continues executing at a high level with second quarter revenues of $23 billion, up 48% year-on-year, ahead of expectations by over $2 billion, driven by strength in GLP-1, which prompted the company to increase its annual revenue and operating margin guidance.

Lastly, we continued to slowly build our position in the fast AI inference provider, Cerebras Systems Inc., after participating in the company’s IPO in the previous quarter. We took advantage of the stock price volatility, which was caused by broader investor concerns over the durability of the AI buildout as well as the limited supply the company currently faces, which has created short-term margin headwinds. We believe that fast token generation will become increasingly important, especially for agentic AI, where many internal tokens are generated for every external, customer-facing token. Speed matters because latency compounds at every step of an agentic workflow; the faster tokens are generated, the more useful and attractive the AI agent becomes. We believe customers will be willing to pay higher prices for faster tokens, as can be evident by the higher price charged by OpenAI for its recently announced, fast-inference service – ‘Ultrafast’, which is priced at a 6x multiple compared to regular speed, for an 8x speed up.

We believe there is a long runway for growth in AI inference in general, and agentic AI in particular underpinned by the continued rapid progress on the pareto frontier of AI intelligence versus price, with a rapid model release cadence, while the cost per a given level of performance is dropping faster than other transformative technologies in history. As agentic AI adoption grows and latency becomes increasingly important, we expect Cerebras’ solution to become more valuable to customers.

“The lesson. Stock prices don't always tell you how businesses are doing. They just tell you what you can buy and sell stock for at a given day and a given moment. Not always accurate. In fact, frequently inaccurate. Our business based upon figuring out what a business will be worth in five or 10 years. And buying at a big discount from that level. That's how come we've been so incredibly successful.”
– Ron Baron in an email to the Firm’s research staff on October 8, 2026.

We have often written in the past that in our view the competitive advantage of successful long-term investors is not rooted in access to better information or in their capacity to perform more sophisticated financial analysis. It is their ability to remain patient through the bouts of market volatility. We think it is not only normal but is a function of human nature. We extrapolate recent trends far into the future and feel invincible near market peaks and helpless near market bottoms. Sometimes, optimism becomes euphoria while fear turns into panic. Volatility can be thought of as a price successful investors pay for the possibility of generating long-term excess returns.

As we are writing this letter, the market is selling off, with the “AI trade” or the perceived beneficiaries of AI getting hit particularly hard. Investors are concerned with the spiking debt yields of neoclouds and recent disclosures by Anthropic that the torrid pace of its growth in revenues and ARR has slowed down. The geopolitical uncertainty remains elevated with the price of oil exceeding $100 once again and the 10-year Treasury yield hovering around 5.3%. There is a broad consensus that the Fed will continue to raise interest rates, which is a headwind to equities and in particular to growth equities, for obvious reasons.

Having said and acknowledged all that… we continue to be incredibly optimistic about the long-term prospects for AI and for the Space Economy. Some years ago, we took a deep dive into NVIDIA and wrote to you how “we have never seen a company like it!” In the last quarterly letter, we took a deep dive into SpaceX (we have owned it for years) and described it as “THE company with N=1!” because in our 30-plus years of professional capital allocation experience we have never seen a company like it. After spending six hours listening to the management of Anthropic at their October investor day and getting a first look at their financials – we can unequivocally state – we have never seen a company like it!

We do not know how or when the wars with Iran and Russia-Ukraine will end, but we believe that they will and that the price of oil will normalize and stabilize at significantly lower levels. We do not know what the Fed is going to do but we do know that technological advancements are deflationary in nature and we believe that progress in AI is real, material, and sustainable, and that it will lead to meaningful improvements in productivity and likely accelerate global GDP growth. More importantly, we believe that the quality of our portfolio companies today is the highest it’s ever been. While corrections, pullbacks, and even bear markets are par for the course, we do NOT see an AI bubble. “Air pockets,” supply constraints, and rising interest rates can all cause stock prices to compress, but in our view, comparisons to the internet bubble of 1999 do not hold water, and even unlike 2021, we do not view our stocks as expensive compared to the strength in their fundamentals.

Each quarter, we review the Fund’s weighted-average valuation multiple and consensus forecasts for 2026 revenues and operating income^9. In the third quarter, the multiple declined by 6.1%, or 2.2% excluding SpaceX. Year to date, the multiple contracted 7.0%, excluding the recent IPOs SpaceX and Cerebras. These declines imply that fundamentals accounted for over 100% of the Fund’s gains for both periods, which often bodes well for prospective returns. Fundamentals also improved as 2026 weighted average revenue forecasts rose 4.8% during the quarter and 12.3% year to date, while operating income forecasts increased 4.2% and 15.5%, respectively. Every GICS sector except for Health Care saw multiple contraction during the quarter, yet revenue forecasts rose in every sector and remarkably, in every sub-industry. Operating income forecasts also increased across all sectors and all but one sub-sector, where lower expectations for Tesla’s operating income weighed on the outlook. We also compared our companies’ current valuation multiples with their five-year averages. At quarter-end, the portfolio’s weighted-average multiple was 12.6% below its five-year average.

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 uncertainty, and taking advantage of the long-term opportunities that those risks and uncertainties create.

We remain optimistic about the long-term prospects of the companies in which we are invested and continue to search for new ideas and investment opportunities. We remain patient, investing only when we believe a company’s price is attractive relative to its intrinsic values.

We thank you for your continued trust and for being our partners on this journey.

Sincerely,