Eaglestone Capital 4Q2025 Investor Letter
January 16, 2026
Dear Partner:
In a volatile 2025, Eaglestone Capital Partners LP ("Eaglestone Partners" or the "Fund") outperformed the S&P 500 by 10%, generating returns for the typical limited partner of 27.53%. Year to date 2026, the portfolio is up 10.45%:
For CY 2025, long positions in Comfort Systems USA (FIX), IES Holdings (IESC) and Interactive Brokers (IBKR) were the largest contributors (in that order) to the Fund's positive performance while long positions in BellRing Brands (BRBR), Coinbase (COIN) and Kinsale Capital (KNSL) were the largest detractors from performance (in that order).
While our long-term thesis on the U.S. industrial base drove these results, the fourth quarter served as a period of consolidation and decisive portfolio pruning. We moved to exit positions where the investment thesis had drifted, accepting short-term volatility to protect long-term capital. This discipline has cleared the decks for a robust start to the new year; year-to-date 2026, the portfolio has already returned 10.45% compared with 1.44% for the S&P 500.
Our Differentiated AI Supply Chain View
Our performance continues to be driven by a highly differentiated view on the Artificial Intelligence supply chain. While the market crowded into digital assets and chip designers, Eaglestone focused on the physical bottleneck — the Engineering, Procurement, and Construction (EPC) firms building the data centers themselves. This thesis carried the day, offsetting the headwinds from our exits in BellRing Brands and Coinbase.
The Evidence: AI Demand Remains "Insatiable"
There's plenty of demand for AI-inference as evidenced by Sensor Tower's report:
- ChatGPT: December 2025 daily active user (DAU) growth was 262% year over year, hitting 392 million mobile DAU
- Gemini: Daily active user growth was 351% year over year, hitting 62 million mobile DAU
Similarly, TSMC's 4Q2025 earnings release showed robust demand for AI-related silicon chips:
- TSMC's high performance compute (HPC) platform (which includes AI accelerators from Nvidia, AMD, etc.) grew 48% year-over-year for the full year of 2025
- TSMC raised their 2026 capex guidance to $52 billion - $56 billion (up significantly from ~$41 billion in 2025)
- Management explicitly stated this spend is to build capacity for 2nm and A16 technologies to meet "insatiable" AI-related demand
- TSMC management, known for being conservative, revised their long-term growth estimates upward, raising the CAGR forecast for AI accelerator revenue to 54% - 56% (2024–2029), up from the previous estimate of 45%
The Bottleneck Has Shifted: Satya Nadella's "Warm Shells" Comment
However, from comments made by Satya Nadella in November 2025, it appears that procuring HPC chips (GPUs) is no longer the main constraint — finding the power and data centers to house them is.
Nadella admitted that Microsoft is "not chip supply constrained" anymore and noted that Microsoft has "chips sitting in inventory" that they cannot currently use. Nadella identified the specific choke point as a lack of "warm shells" — industry jargon for data center buildings that are fully constructed and have power and cooling connected, ready for server racks to be rolled in.
This commentary aligns with the broader industry trend where the limitation is moving from the "digital" supply chain (wafers/chips) to the "physical" supply chain (transformers, concrete, and electricity).
Why FIX Continues to Win
This commentary supports Eaglestone Capital's view that the large, efficient and experienced contractors (like FIX and IESC) who can help the hyper-scalers procure materials and equipment for the building of data centers while employing modular construction techniques to help speed up the construction process will continue to be in demand as the AI-capex buildout rolls on.
FIX's "Off-Site Manufacturing" (OSM) Strategy
FIX offers an "Off-Site Manufacturing" (OSM) strategy that moves the complexity of data center construction from the field to the factory. This is their primary solution for speed.
For example, instead of building cooling systems pipe-by-pipe on the construction site, FIX builds massive, fully integrated modular central plants and mechanical/electrical rooms in their own factories (specifically via their subsidiaries like EAS and TAS Energy).
Parallel Processing: While the data center's concrete shell is being poured on-site, FIX is simultaneously building the cooling and power infrastructure in a factory in Texas or North Carolina.
Specific Modules:
- Cooling Skids: Pre-fabricated frames containing pumps, chillers, and heat exchangers
- Electrical Rooms: Modular units housing switchgear and power distribution
- Multi-Story Assemblies: They can stack these modules to create multi-story mechanical penthouses that are simply trucked to the site and lifted into place
The Speed Gain: Reports indicate this method can reduce construction schedules by up to 40%, directly addressing the "warm shell" delay Satya Nadella mentioned.
FIX's Procurement Advantage
In addition, FIX plays a major role in procurement, effectively acting as a supply chain buffer for the hyper-scalers.
Buying Power: Because they are one of the largest mechanical contractors in the U.S. (with a backlog over $8 billion as of late 2025), they have massive leverage with equipment manufacturers.
Securing "Long-Lead" Items: The "essential materials" in this context are not just steel and concrete, but the hard-to-get HVAC chillers, massive water pumps, and electrical switchgear. FIX procures these months in advance to build into their modules.
Inventory Management: By taking delivery of these components at their own factories (rather than waiting for the data center site to be ready), they prevent supply chain delays from stalling the project.
The AI-Tax Problem for SaaS Companies
Given the carnage in software, Eaglestone is still looking for good companies to invest in but very few seem to have "figured it out."
The problem for software as a service (SaaS) companies looking to capitalize on the perceived value of AI-intelligence is that AI software introduces a fundamental operating cost that never used to exist. AI is NOT free.
Unlike the last decade where a SaaS company could invest in the upfront costs of building a product with a beautiful and intuitive UI, cleverly coded to be fast with low-latency and then set about adding users with a near zero incremental cost to serve, the AI-era now imposes a marginal inference cost far higher than zero, so each additional user adds to their costs.
In fact, costs are driven by usage, hence their "best clients" who use the AI-infused software the most, are also their largest cost.
Palantir: The Exception
Whether through luck or foresight, Palantir built a business model focused on deploying their engineers in-house to holistically re-architect a client's software stack to hit performance targets and be paid only after those targets were hit, while the client paid for the inference costs. That has worked out very well for them and now others are trying to transition to the "performance model" from the traditional SaaS fixed monthly fee per seat model where the SaaS company covers the running costs.
Like the time and pain endured by MSFT, ADBE and others to transition from an upfront purchase model for software to a monthly subscription model, we are now witnessing yet another business transition to an AI-software model that will reward firms based on outcomes. The transition is painful, yet many are trying it out and perhaps they will succeed.
Consequently, we are avoiding the "SaaS middle-class" where inference costs will erode margins. Eaglestone is looking for companies that can demonstrate AI-related revenue growth with margin expanding operating leverage — but apart from Palantir, Eaglestone is still looking.
BellRing Brands: Final Post-Mortem
Following a strong performance in FY2024, we admit to being overly confident in BellRing Brands (BRBR) prospects. The basic thesis behind Eaglestone's investment in BRBR was a derivative of our belief that Americans prefer taking a "pill" to cure their weight-related issues as opposed to hitting the gym.
Yes, GLP-1 drugs are real and their adoption is growing. Some researchers claim that over 20% of US households have at least one member who is taking one of these weight loss drugs. The issue is how to find investments that will benefit from this trend?
BRBR has been a beneficiary, as GLP-1 users typically suffer from loss of muscle mass and hence they are encouraged to take fast and easy high-protein foods like BRBR's Premier Protein ready to drink shake.
As discussed in previous letters, the deeper issue lies with the strength of Costco's and Walmart's distribution channels and BRBR's focus on taste over performance, which has overwhelmed BRBR's business and turned them into a "white label" product for those powerful retailers, lowering sales velocity and margins.
The position was closed out in early August 2025.
Coinbase: A Detailed Analysis
COIN is interesting because their partnership with Circle (CRCL) grants them most of the economics for the USDC stablecoin.
The USDC Economics
USDC reserves (the stablecoin is backed 1:1 by primarily US treasuries) have grown over 56% during the past 12 months as usage has increased and now total over $75 billion which yielded COIN $354.7 million in 3Q2025 revenues (or $242 million after marketing rewards paid by COIN generating a 68% operating margin).
USDC is primarily designed to enhance transaction efficiency for payments — the yield on the reserves is paid to Circle and COIN, not the holders of USDC.
The Revenue Split:
- For USDC held within Coinbase's own products (~$15 billion): Coinbase retains 100% of the interest income (COIN earned an effective 1.25% yield after rewards payments)
- For USDC held outside of Coinbase (~$53 billion): Interest income is split 50/50 with Circle (COIN effectively earned a 1.5% yield)
- The remaining ~$5.5 billion was held on platform by CRCL, who received 100% of that yield
The Bull Case
There are strong arguments to justify continued growth in USDC's usage and hence reserves. USDC allows for instant low-cost transfers and therefore has the potential to disrupt swipe fees, corporate float costs, bank ACH/wire fees and FX payments/remittances.
Some forecast that USDC's reserves will hit $2 trillion to $4 trillion over the next decade which would represent a CAGR of between 39% to 49%.
Assuming a more conservative 25% CAGR over the next 10 years, USDC reserves would grow to $700 billion. Assuming the existing relationship between Circle and COIN remains, this would imply a present value of COIN's USDC-related operating income to be worth between $80 billion to $90 billion discounted back to today.
The Valuation Puzzle
Given that COIN currently trades with a market capitalization of about $65 billion you could argue that you're getting the rest of COIN's business for free with a buffer to cover you for their $15 billion in liabilities.
Indeed, apart from the Base L2 network that has become the dominant Ethereum layer-2, it's unclear to me that COIN's remaining businesses (the core exchange, the Deribit derivatives business, the custody business and the staking business) have sufficient defensive moats to protect them from the onslaught of competition in the crypto space and we fear that these other businesses might overwhelm their growing and attractive USDC business.
Why We Sold
COIN's operating income is still very volatile, highly geared to crypto trading flows. As further evidence, COIN's stock price has a beta of 3.7 versus Bitcoin (so for every 1% change in BTC, their stock moves 3.7%).
COIN is interesting but the risks in their non-USDC business give me a lot to worry about and we closed out the position in November 2025.
While the "Sum of the Parts" math suggests Coinbase is undervalued — implying the exchange business is free — the market is heavily discounting the durability of that revenue stream due to regulatory overhang and commoditization risks.
We see the logic in the long thesis, but at Eaglestone, we prefer "inevitability" over "possibility."
Regards,
Eaglestone Capital Management LLC
66 Palmer Avenue, Suite 32B, Bronxville, NY 10708
914.202.8811
eaglestone-capital.com
Eaglestone Capital Management LLC prepared this letter. NAV Fund Services, our administrator, is responsible for the distribution of this information and not its content.
Eaglestone Capital's Investment Philosophy
Eaglestone Capital is an independent registered investment adviser (RIA) based in Bronxville, NY, founded by Fred Stupart. It serves as the sole investment adviser to Eaglestone Capital Partners LP (the "Fund"), which targets 15% absolute annual returns to its limited partners through investments in daily-liquid securities. The Fund primarily invests in U.S. equities and may employ leverage to boost returns.
Eaglestone Capital views the U.S. economy as sui generis — truly one of a kind: unparalleled, irreplaceable, and impossible to replicate.
The foundation of continued U.S. economic growth — encompassing employment gains, real estate appreciation, and other asset expansion — is its dynamic, highly competitive private sector, exemplified by many publicly traded firms. As a result, investing in high-quality U.S. companies that uphold shareholder rights is vital to any robust investment strategy.
Eaglestone Capital believes that intelligence, effort, and patience yield the best results. While this should be self-evident, many index fund philosophies conflict with it. When Jack Bogle introduced the S&P 500 index fund in 1976, he offered a new, low-cost product that helped keep large asset managers honest. Yet what is initially a clever innovation often becomes overused. We may soon see many index funds lag behind AI-empowered active managers able to allocate capital more flexibly.
Eaglestone Capital oversees a portfolio of 20–30 liquid equities, each run by leadership teams possessing a "fiduciary gene" — managers who respect shareholder rights and strive to develop innovative, capital-efficient businesses that deliver strong returns to owners.
Finally, Eaglestone Capital does not equate volatility with risk — in fact, volatility is a key benefit available only to public market investors seeking high absolute returns. Sometimes "the market" decides that it wants to sell stocks — Eaglestone Capital respects the market's momentum, appreciates its power and will wait for an appropriate time to exercise any judgement.
General Disclaimer: This letter and its contents are confidential and proprietary. Past performance is not a guarantee of future results. Full disclaimer available upon request.