The purpose of markets is to price risk and reward, thereby helping the capitalist system allocate resources intelligently. This “invisible hand” works strikingly well. Areas of the economy with promising or expanding profit pools attract investment and talent, while struggling ones shed both. Investors often face the unpleasant dilemma of shifting capital between struggling but cheaper businesses and promising but more expensive ones. Investing in a broad index like the S&P 500 allows passive investors to participate in the positive drift of equity markets without having to navigate these trade-offs.
Occasionally, there is such a compelling theme that it attracts capital to the point that unrelated businesses, even those with excellent prospects, are left by the wayside. Increasingly, we think the AI capex boom is one of these moments.
We are enormous believers in AI. We are even stronger believers in building a portfolio for a range of possible futures, rather than confidently predicting a specific one. Many investments in the AI capex supply chain are pricing optimistic cases without sufficient consideration of downside risks. Because the theme is now a large part of the overall stock market, the S&P 500 is increasingly exposed to this risk. The market’s high valuation also reduces expected future returns.
By contrast, one can build a portfolio with a broad range of bets that looks likely to generate attractive long-term returns, without undue risk. Eagle has been methodically recycling capital to pursue this. We still have exposure to AI capex investments, albeit less than before. We also have some investments with potential for AI disruption where we think the risks are limited and well-priced. However, most of the portfolio is in high-quality companies where AI is, at most, incidental to the investment thesis.
The AI capex boom is driving extraordinary growth for the S&P 500, with recent earnings up almost 20% year-on-year. While this is exciting, it is also buoyed by a transitory accounting effect. When a semiconductor company sells chips or racks, it books 100% of the revenue on the day of the transaction, and the buyer records zero expense. The equipment is recorded on the balance sheet and expensed over 5-6 years through depreciation. The initial earnings impact for the market is all dessert and no spinach.
By contrast, free cash flow, which captures this effect, is growing only in the low single digits. OpenAI and SpaceX, along with many smaller VC-backed firms, are loss-making, and a significant portion of their expenses are reflected in the S&P 500’s revenue. Without that subsidy, free cash flow growth for the S&P 500 would be even lower.
This is not to say investments in AI are bad. To state the obvious, AI is an incredible technology that will have profound effects on the economy and the world. Most capex is currently generating good returns, including for companies Eagle owns. Still, we bear in mind that the positive impact on earnings is a mechanical effect that happens early in every capex boom. At best, earnings growth gently fades as the higher level of investment is absorbed through the economy. At worst, investors over-extrapolate the demand curve and underestimate supply-side adaptation, leading to a collapse in capex and earnings.
Our base case is that demand for AI-enabling technology will remain robust. Shortages in capacity to deliver AI capex are likely to persist in the near term. We also expect benefits to diffuse through the economy and broaden beyond the picks and shovels delivering the buildout. We anticipate heightened creative destruction, as white-collar workflows evolve and competitive moats erode—creating, inevitably, winners and losers.
One of the ironies of today’s environment is the level of disruption AI is expected to drive across many downstream businesses, compared to complacency about the disruption that may later negatively impact current beneficiaries. Capitalism is superb at chasing outsized economic returns and competing them down. The very success that AI picks and shovels are enjoying today risks damaging the attractive industry structure within which they exist.
Among the model companies, OpenAI and Anthropic are clearly out in front. OpenAI has the leading consumer chatbot, and Anthropic’s Claude Code took the enterprise world by storm this year. The growth that Anthropic has seen is unprecedented, going from tiny to over $60 billion in run-rate revenue over the past twelve months.
The success of these companies has led to valuations of approximately $1 trillion each, with expectations that one or both may trade near $2 trillion post-IPO later this year. Their growth and high valuation have enabled them to invest heavily in compute commitments with hyperscalers, as investors see strong and improving “inference margins,” which measure profitability excluding the compute costs of training models, R&D, sales & marketing, and other overhead.
If there were to remain only two companies capable of delivering state-of-the-art models, the AI labs would be excellent businesses. However, Alphabet, SpaceX, and Meta all have enormous model investment programs. At times, Alphabet has been ahead. Today it’s not, but with each new release, the pole position shifts. SpaceX and Meta were well behind and have recently made significant strides, offering increasingly competitive models at a far lower price point. In the background, open source continues to improve, narrowing the performance gap.
If we end up with 4-5 leading companies, plus open source, a weaker industry structure will emerge. This is particularly the case since some of the companies have other large businesses and ways to monetize the value chain, even if they never make money on the models themselves. Moreover, as AI tokens become a larger expense item for corporate America and the world, there will be pressure to optimize them and improve their efficiency. It is likely that a small percentage of use cases will require an expensive state-of-the-art model; most will be routed to more efficient tools at a fraction of the cost. It is too early to judge, but predictions that Anthropic will go from $60 billion in revenue to over $1 trillion in the coming years run counter to most of technological history.
Amazon Web Services (“AWS”), Microsoft Azure, and Google Cloud Platform (“GCP”) are highly profitable businesses with strong growth, margins, and returns on capital. We own positions in each. These businesses are growing faster and will be even bigger than we previously thought. However, there is little question that the industry structure they inhabit is worse than it was a few years ago.
Until recently, cloud infrastructure was a 2.5 player market: AWS and Azure dominated, with GCP a distant third. Today, GCP is highly competitive, and Oracle, SpaceX, and a myriad of neoclouds have come into the mix. Perhaps even more importantly, the industry now faces customer concentration risk unlike any it saw before. OpenAI and Anthropic account for a large share of these companies’ backlogs and growth. To the extent this trend continues, it is likely to pressure the returns the hyperscalers can earn. OpenAI and Anthropic will be large and sophisticated enough to bid hyperscalers against each other or even to build their own data centers.
Despite this deterioration in industry structure, we think the hyperscaler business model remains reasonably strong. AWS, Azure, and GCP have powerful go-to-market and ecosystem advantages that give them superior value propositions compared to neoclouds. Additionally, each of them has internal silicon programs that should give them cost advantages. We think these companies offer attractive ways to participate in the AI compute toll-road without undue risk. We particularly like Amazon, which has the broadest business, well-developed internal silicon programs, a history of operating as a low-cost producer, and a stock that has largely sat out the AI bull market.
The semiconductor value chain has benefitted mightily from the AI capex build-out. It is hard to find losers. Over the past five years, the SMH index has compounded at 39% annually compared to 13% for the S&P 500. Semis now account for almost 20% of the S&P 500, with power infrastructure and other parts of the broader AI supply chain pushing this up to more than 30%.
While near-term earnings are likely to remain strong given widespread shortages, long-term risks are mounting. There are parallels to the China commodities boom of the early 2000s. As with AI, China’s strong demand after it entered the WTO caused widespread shortages, explosive earnings, optically cheap multiples, and predictions of permanent undersupply. Then, demand moderated, commodities production was overbuilt, and prices and profits collapsed. Historically, one of the most attractive characteristics of the semiconductor supply chain was the degree of consolidation across many sub-markets. This industry structure is changing and now faces new pressures.
NVIDIA has dominated the market for AI accelerator chips and has become the most profitable company in the U.S. However, its largest customers have internal silicon programs that are likely to reduce NVIDIA’s share over time. Google TPUs and Amazon’s Trainium chips are the most competitive, but Microsoft, Meta, OpenAI, and Anthropic all have programs in various stages. When the market was small, custom programs were uneconomic. At today’s scale, R&D can be amortized across the larger base. Moreover, NVIDIA’s 75-80% gross margin implies it is marking up Taiwan Semiconductor (“TSMC”) wafers by 4-5x. Custom programs don’t have to be as good as NVIDIA to offer better price/performance to the end customer.
NVIDIA is investing in the neocloud industry to support demand for its chips and weaken hyperscalers’ position. Ironically, by selling its chips at high prices, it makes the neoclouds high-cost operators, and therefore less likely to succeed against vertically integrated hyperscalers over the long term. As NVIDIA’s market share crests and begins to fall, margins may follow as it faces both direct and indirect competition from its largest customers.
Micron, the only U.S. producer of DRAM memory, has gone from being marginally profitable to earning more than Apple or Microsoft. Memory prices are now high enough to inflict damage on consumer electronics markets and prompt AI model companies and chipmakers to redesign their technology.
Micron’s entire manufacturing PP&E footprint is approximately $60 billion, versus a market cap of $1.2 trillion. The company will make enough money over the next year to replicate its entire footprint multiple times. DRAM historically has been brutally cyclical. That cyclicality reduced competition to the point that only an oligopoly of three players could survive. Volatility was one of the moats that helped protect over-the-cycle returns.
Today’s extraordinary profits and prices are going to kick off an equally extraordinary capex cycle. Micron, along with Samsung and SK Hynix, is beginning to add capacity, though too slowly to keep up with demand. Chinese DRAM producer CXMT is a technology laggard that has struggled to become competitive. However, given the strong markets, we expect CXMT to add more capacity than any incumbent. Before long, DRAM is likely to transition to a four-player market, and the Chinese may become the largest producer. Historically, this type of change hasn’t been good for industry profits. NAND memory, used for longer-term storage, is likely to fare worse, as Chinese producer YMTC is further along in this sub-sector.
Even firms like TSMC face increased risk. TSMC management’s understandable reluctance to overinvest in capacity has caused shortages that created opportunities for Intel and Samsung. While we think TSMC’s technology leadership position and comparatively mild pricing will mitigate the pressures, it is likely to see tougher competition than we would have expected a couple of years ago.
Taking a step back, the semiconductor industry has long been a productivity force that creates value in part by driving down prices. Recently, this trend has gotten off track. Extraordinary price increases in the memory industry, exceeding 7x, exemplify this dynamic. Semiconductor products complement each other. If memory costs more per bit, the customer can’t spend as much on other parts of the value chain. Impacted suppliers, many of whom also serve as choke points, may charge higher prices to protect their share of industry profits. When semicap firms see their customers’ prices and margins escalate, they too raise prices. Inflation begins to cascade widely through the supply chain, contributing to the cycle’s own undoing.
***
Fund flows chasing the capex theme are creating attractive opportunities elsewhere. The broad equity market seems riskier than usual yet offers pockets of unusual value in some of the world’s best companies. We have added several as new positions in recent quarters.
We believe that, compared to the overall market, Eagle’s portfolio has higher-quality businesses, trades at a 20% discount, and has faster EPS growth—widening the discount over time. We estimate Eagle’s portfolio trades at under 15x 2027 earnings and that EPS will grow in excess of 15% in the coming years. On the following pages, we provide an overview.
London Stock Exchange, Danaher, S&P Global, Mastercard, A.J. Gallagher
Approximately a quarter of Eagle’s portfolio is spread across a handful of superb businesses that exhibit high and stable margins, strong returns on capital, wide competitive advantages, and well-above GDP growth. We expect the group to deftly navigate the integration of AI into their markets. Most operate with a combination of oligopoly/monopoly assets, network effects, or regulatory moats.
London Stock Exchange owns a mix of financial market infrastructure, including the London Clearing House, Russell Indexes, Tradeweb, and Refinitiv data and analytics. Danaher is a leading life sciences company, selling a broad mix of consumables and tooling for biological R&D and drug production. S&P Global owns the S&P ratings franchise, S&P Indexes, Platts, and Capital IQ. Mastercard operates as part of a global duopoly of payment rails and standards. A.J. Gallagher is a commercial insurance broker that continues to gain share in a consolidating industry.
A few years ago, these stocks were priced for perfection. Their attractive characteristics were prized, and the group traded at more than 30x earnings. Over the past five years, the businesses have grown significantly, while the stocks have de-rated. Today, the group trades at a high teen multiple. In a market that is growing EPS at nearly 20%, their steady growth isn’t scarce. However, as earnings in the economy normalize back towards mid-single-digit nominal GDP growth, these businesses should demonstrate significant outperformance.
We expect EPS growth to approach mid-teens over the next several years.
TSMC, Amazon (60%), Microsoft (50%), Alphabet (35%), ASML
Half of our capital here is invested in TSMC and ASML, two of the most dominant businesses in the semiconductor supply chain. The other half is invested across the three major hyperscalers. We ascribe a portion of the value of each of these three companies to their cloud business and the balance to the remainder of the company; i.e., 60% of the value of Amazon is from AWS and 40% is from retail.
TSMC and ASML are two of the best businesses in the semiconductor industry. TSMC is the world’s fab, where most leading-edge logic chips, ranging from Apple to NVIDIA, are made. ASML has a monopoly in extreme ultraviolet lithography (“EUV”), which is required for leading-edge semiconductor manufacturing. Both companies are well positioned over the next 5-10 years to participate in the sector’s secular growth while maintaining their competitive positions. That said, they will face pressures when the next downturn comes.
As previously discussed, we think the hyperscalers have wide competitive advantages over neoclouds. Earnings have accelerated as the companies have significantly increased data center capex. Our central case is that they will provide control-plane solutions that let customers choose among AI models and capabilities, and that AI will diffuse broadly enough to mitigate, but not eliminate, customer concentration risk with Anthropic and OpenAI. Returns on capital should still be attractive, albeit not as high as before. Earnings growth and earnings should be much higher in the coming years given the magnitude of the investment and opportunity.
We expect EPS growth of 15-20% over the next several years.
MercadoLibre, Amazon (40%), Alphabet (65%), Meta Platforms
Operating massive consumer platforms that aggregate demand for sellers and advertisers, these companies have scale advantages, fast growth, and the ability to deploy AI to further press their leads.
Almost half of our capital is in MercadoLibre, the leading Latin American e-commerce and fintech company. Latin America’s relatively low e-commerce penetration—we estimate it at the mid-teens compared with nearly 30% in the U.S.—offers a long runway for growth. In recent quarters, revenue grew more than 40%. We think the market is too focused on near-term earnings revisions and may be conflating elective investments with structural weakness. Much like Amazon in its early days, MercadoLibre is trading lower margins today for a larger business tomorrow; we’re thrilled with this approach, which both builds value and has enabled Eagle to establish an attractively priced position.
Amazon’s retail business, Google Search and YouTube, and Meta are comparatively more mature, but they continue to grow at well-above GDP rates with attractive margins. We believe AI is demonstrably helping Google Search and Meta’s advertising business, both of which have accelerated as ad targeting and content capabilities have improved. Amazon and MercadoLibre should also benefit more than brick-and-mortar peers, which are unlikely to capture the same benefits from advertising, improvements in consumer search, or gains in warehouse robotic technology.
We expect EPS growth of more than 20% over the next several years.
UnitedHealth Group, Humana, Elevance
These managed care companies have significant scale advantages in a consolidated industry that outgrows the overall economy. The industry moves with its own cycle and, over the last several years, has faced cost/price pressures in Medicare Advantage and Medicaid. Even the largest firms are earning poor margins, and some weaker firms are unprofitable. We believe conditions have bottomed out and that we are transitioning to a multi-year improvement in margins and earnings. AI-driven technology projects can both reduce costs and improve service quality.
Since returning as CEO at UnitedHealth last year, Stephen Hemsley and the management team have been correcting some of the company’s missteps. Recent results provide growing evidence that the turnaround is well on track. Humana is also making good strides, and we expect significantly improved Star ratings for the company later this year.
We expect EPS growth of 15-20% over the next several years, driven by moderate revenue growth and significantly expanding margins.
Workday, SAP, Intuit, Microsoft (50%)
Software is controversial due to fears of AI-driven disruption. AI makes it easier to build software and will change workflows in how it is used. We believe there will be heightened competition and greater separation between winners and losers over the coming years. The industry is deservedly trading at a higher risk premium, but within the market there are plenty of mispricings.
The companies we own are resilient and highly entrenched in application software. We view them as offering upside optionality. Workday, a leading HCM and financials platform, earns mid-teen margins on a business that should have long-term margins of 30% or higher. The founder’s return to the CEO role is reigniting product innovation, and we increasingly believe AI will be a net positive for their business, enabling them to play offense. Examples include geographic expansion, adjacent-product extensions, and natural-language querying for business results.
SAP is the world’s largest provider of ERP software. It is mission-critical and sticky. It will face tougher competition in peripheral products, but its core should grow rapidly over the next five years as it migrates customers to its modern platform.
Intuit’s QuickBooks, the dominant bookkeeping software for small and midsize businesses in the U.S., exists in a market that is difficult to serve profitably, given the low average selling price. Intuit’s brand, scale, and network effect through the accountant channel give it formidable competitive advantages. It is well-placed to deliver AI solutions to this hard-to-reach customer. TurboTax, a relatively smaller part of the business, will likely adapt its business model to an AI world, but we have more modest expectations for its growth.
Microsoft has formidable scale and distribution advantages. It should be well-positioned to deploy AI capabilities across productivity software throughout the economy. However, it is also vulnerable to changing white-collar workflows, and its execution has been mixed, with products like Copilot lagging Claude’s plug-ins. That said, Microsoft has a rich history of being out-innovated by new products and then playing fast follower as it bundles each capability into its platform.
We expect EPS growth in the high teens, driven by revenue growth, operating leverage, and share buybacks.
ConocoPhillips, EQT, Alcoa
The energy & metals companies we own benefit from favorable multiyear supply/demand outlooks, management teams that are good capital allocators, and asset bases that are well-positioned on the global cost curve.
ConocoPhillips, the largest U.S. pure-play upstream oil producer, has low-cost, long-lived inventory that we believe is underappreciated relative to peers. In the coming years, its free cash flow should inflect higher as the Willow project in Alaska comes online and transitions from a cash drag to a cash generator. We see the oil market as reasonably balanced, but risks may skew to the upside over the next couple of years as the world manages shortages from the Strait of Hormuz, and the path to its full and enduring reopening remains murky.
EQT is the largest U.S. pure-play natural gas producer. Like ConocoPhillips, it has low-cost, long-lived assets and management with an excellent track record allocating capital. Its position in the Marcellus shale and internally owned pipeline assets give it a distinctive position compared to most peers. It earns good margins even when prices are low. In the coming years, as LNG export capacity is added and U.S. electricity demand increases due to datacenter buildouts, we think supply and demand are likely to tighten. Combined with moderate production growth and a large free cash flow yield, it offers good EPS growth and attractive skew.
Alcoa is a global producer of aluminum and alumina. For more than two decades, Chinese aluminum supply was relentless, swamping the market. The country has imposed smelter caps and is no longer adding net capacity. As global demand grows, new capacity must be added elsewhere at higher build costs. Incentivizing this capacity requires higher prices to earn a return. We own Alcoa at a wide discount to replacement cost, and this shift higher in aluminum and alumina prices is a powerful tailwind for earnings power.
We expect EPS growth in the mid-teens over the next several years.
Capital One Financial, Lennar, Estée Lauder, Charter Communications
The consumer cyclicals and staples companies in this group have idiosyncratic opportunities to drive rapid earnings growth over the coming years. Capital One, one of the largest credit card companies in the U.S., operates the only full-suite scaled digital bank. Last year’s acquisition of Discover conferred both significant financial synergies and the upside of operating its own network. Despite its size, it is still run by the founder and retains an entrepreneurial spirit. With 20% returns on equity, we believe it is underappreciated and can drive both strong earnings growth and multiple expansion.
Lennar is the second-largest homebuilder in the U.S. Margins and earnings have collapsed over the past few years, driven by weakness in the company’s geographic markets. This has been exacerbated by a change in business model, whereby the company spun off its land bank. In the long-term, going “land light” should reduce balance sheet risk and improve returns on capital, but it also reduces margins, which has a more pronounced effect on earnings in today’s weak market. Over time, we expect excess inventory to clear and volume and discounting to improve. Indicative of depressed valuations, M&A activity in the sector is increasing. Berkshire Hathaway recently acquired Taylor Morrison, making it the fourth-largest builder in the country.
Charter Communications will become the largest cable operator in the U.S. when its acquisition of Cox closes. Cable has been a disappointing area, with heavy competitive pressure and minimal growth. We see Charter equity as a cheaply priced option. 80% of the capital structure is low-cost long-term debt. With a free cash flow yield that will exceed 30% in a couple of years, the company is enormously liquid and can both de-lever and return cash to shareholders. In time, we expect the company to be consolidated by Comcast, a telco, or even a non-traditional buyer. The equity could be up by a multiple in certain scenarios.
We expect EPS growth of approximately 20% for this group of companies over the coming years.
***
Success over the long term requires a willingness to be different in the short term. The two are inextricably linked, and part of the price one pays for superior results. The price can be high. Many peers we admire have struggled over the past 5-10 years. Firms have lost assets or even closed. Some are capitulating to momentum because they feel business pressure. Accordingly, we believe Eagle’s competitive differentiation is widening.
Our goal is to use this differentiation to build a distinctive portfolio that offers higher expected returns while taking less risk. The current environment is unusually fruitful for doing this. Thank you for your steadfast support in pursuing these opportunities.
As always, if you have any questions or would like to discuss anything herein, please call us at (212) 293-4040. Also, if your financial situation or investment objectives have changed, if your IPO eligibility or U.S. Person status has changed, or if you would like to modify or discuss any investment restrictions or guidelines, please reach out to your contact on the client team or email ClientServices@eaglecap.com.1
1 Eagle Equity Composite holdings during Q2 2026 were as follows: AA; AER; AJG; AMZN; AON; ASML; BAYRY; CHTR; CMCSA; COF; COP; DHR; EL; ELV; EQT; FICO; GEV; GOOG/L; HLT; HUM; INTU; LBRDA/K; LEN; LSEGY; MA; MELI; META; MSFT; NFLX; PTC; SAP; SHEL; SPGI; TSM; UNH; VMC; VSNT; WDAY; WFC; and WWD.
Past performance is not indicative of future results, and there is no assurance that Eagle Capital’s investment objectives will be achieved or that the strategies employed by Eagle Capital will be successful. Except where otherwise indicated, the information contained in this content is based on matters as they exist as of the date of preparation of such material and not as of the date of distribution or any future date. More specifically, any content in relation to financial metrics or portfolio positions is as of 06.30.2026. This document does not constitute an offering of advisory services or advice or a recommendation or offer to sell or a solicitation to deal in any security or financial product in any jurisdiction. It is provided for informational purposes only and on the understanding that the recipient has sufficient knowledge and experience to be able to understand and make their own evaluation of the content described herein, any risks associated therewith and any related legal, tax or other material considerations. Recipients should not rely on this material in making any future investment decision. To the extent that the reader has any questions regarding the applicability of any specific issue discussed above to their specific portfolio or situation, clients and prospective investors are encouraged to contact Eagle Capital as well as to review other materials produced by Eagle Capital (including those available at eaglecap.com). Discussions herein relating to risk or any efforts to mitigate risk do not imply that any actions taken by or investment strategies employed by Eagle Capital are in any way low risk or risk free.
This document expresses the views of Eagle Capital as of the date indicated and such views are subject to change without further notice. Eagle Capital has no duty or obligation to update the information contained herein. Certain information contained in this content constitutes “forward-looking statements,” which can be identified by the use of forward-looking terminology such as “may,” “will,” “should,” “expect,” “anticipate,” “project,” “estimate,” “intend,” “continue,” or “believe,” the negatives thereof, other variations thereof or other comparable terminology. Due to various potential risks, uncertainties or actual events, the results or the actual performance of Eagle Capital’s investments may differ materially from those reflected or contemplated in such forward-looking statements.
Any discussion of specific companies contained herein is for informational purposes only and does not represent all of the securities purchased, sold or recommended by Eagle Capital. Additionally, certain companies that are referenced herein are not held in the Eagle portfolio; a list of all Eagle Equity Composite holdings for the applicable period appears above. The reader should not assume that any investments in the securities identified and discussed herein were or will be profitable.
Any index referenced herein is presented because Eagle Capital feels that it serves as a useful point of comparison with aspects of Eagle Capital’s portfolio management. The S&P 500 Index includes approximately 500 leading companies and captures more than three quarters of the total market capitalization. It is float-adjusted and based on the market cap weightings of the securities that comprise the index. In contrast, the Eagle Equity Strategy is highly concentrated and may contain companies not listed in the S&P 500 Index.
Q2 2026 Letter: One Theme, Many Futures | Eagle Capital Management
The AI capex boom now dominates the market, pricing in a specific optimistic scenario and leaving excellent businesses by the wayside.
