Hosking Partners is a London boutique founded in 2013 by Jeremy Hosking, a portfolio manager at Marathon Asset Management for over 25 years. It runs a single global equity strategy built on the capital-cycle, supply-side approach — contrarian, long-term, and unusually diversified (350+ holdings) under a multi-counsellor model, managing around $5.5bn.
This report says the fund manager is cautious about the AI infrastructure boom, warning it could become commoditized like solar power. The fund returned +3.5% this quarter and +19.3% year-to-date, beating its benchmark. Instead of chasing big US AI stocks, they picked winners from overlooked places like Norway, Greece, and the UK. Key holdings include Saga (a UK cruise company with a 25% annualized return, similar to Meta and Microsoft) and ASOS (a turnaround story). They also bought office properties Vornado (New York) and Great Portland Estates (London), citing limited supply. The overall tone is cautious on AI hype.
One-sentence summary: The author adopts a cautious stance on the AI infrastructure investment boom, warning it may repeat the commoditization trajectory of the solar industry, while emphasizing that a diversified stock-picking strategy and uncovering winners from overlooked corners of the market are proving effective. [Cautious]
Performance Comparison: Net return of +3.5% for the quarter and +19.3% year-to-date, outperforming the MSCI ACWI Index by 1.9% and 6.3%, respectively. Year-to-date 2026, the strategy has outperformed the benchmark over 1/3/5/10-year periods, achieved against a backdrop of an increasingly narrow AI-dominated index and the US 10-year Treasury yield rising to its highest level since 2007. The author believes this reflects alpha generated from a long-term, uncorrelated, and idiosyncratic stock-picking approach.
The author argues that AI infrastructure investment has entered a cycle "beyond all hyperbole," making three observations:
The author reiterates their capital cycle framework: if the large language model market remains competitive and relatively commoditized, returns will trend toward marginal cost. It is now increasingly clear that data centers are factories, and compute is a commodity. In competitive, asset-heavy industries, 40 years of experience show that the "urge to run at full capacity" will drive behavior—to cover high fixed costs, pricing (i.e., compute tokens) will decline, squeezing returns for all participants. The author proposes an analogy with significant implications for global equity markets: AI data centers may follow the same path as the solar industry—a revolutionary technology rapidly adopted, ultimately commoditized, and dominated by China. The author states: "With a supposed $3.5 trillion combined IPO valuation, this conundrum will be tested by the progress of OpenAI and Anthropic towards their flotations in the upcoming months."
Performance Comparison: The top ten contributors were not large US AI-themed beneficiaries but came from countries like Norway, Greece, South Africa, and the UK, spanning sectors including Energy, Financials, Materials, and Consumer Discretionary, with a median market capitalization of just $4.7 billion. Many positions have been held for several years, achieving multiple returns. The author uses this to challenge the standard interpretation of Bessembinder's research—that one should focus on predicting and holding a few "winning" stocks.
This quarter's returns were driven by three structural factors, which would be amplified if the AI cycle reverses. First is successful stock selection—in any market with broad dispersion, stock picking yields high returns. Second is the structural advantage of Hosking Partners' global, all-sector, generalist approach: not being tied to a specific investment style (such as "quality" or "value") or narrow sectors/geographies allows free comparison and exploitation of opportunities anywhere. The author cites an example: "The ability to compare and act on a cheap, near-monopoly Greek refiner versus an expensive, highly leveraged US enterprise infrastructure stock is rare and valuable." Third, the tolerance for large positions enhances the capacity to invest in areas with little or no competition. This ability to bear uncorrelated, idiosyncratic "risk" was central to many of the quarter's major performance contributors, and the author believes it will remain so in the years ahead.
The author cites academic research, arguing that market winners are difficult to predict and are more likely to emerge from overlooked corners than from current consensus hotspots. Citing Bessembinder's research, the author notes that over the long term, a small number of stocks have generated the bulk of net wealth in the US stock market. The author concludes: "We believe winners are hard to predict, and they are as likely to emerge from neglected corners of the market as from today's consensus favorites." This echoes the report's cautious stance on the AI infrastructure investment frenzy—current consensus is heavily concentrated in a handful of AI-related stocks, while the true long-term winners may come from other areas.
| Ticker | Direction | Author's One-Sentence View | Key Data |
|---|---|---|---|
| Saga | Hold & Observe | ROIC at 25%, stock up 11x from 2022 lows, still has upside | ROIC 25% (annualized), NatWest joint venture bank absorbed £2.1 billion in deposits over 9 months |
| ASOS | Hold & Observe | Cautiously optimistic about Saga-like upside potential | Management is executing a turnaround plan years in the making |
| Vornado | New/Add Position | Favorably positioned due to supply constraints | Provider of high-quality office buildings in New York |
| Great Portland Estates | New/Add Position | Favorably positioned due to supply constraints | Provider of high-quality office buildings in London |
| Aichi Financial Group | Hold & Observe | One of the top ten contributors in Japan, steady performance | Japanese financial sector |
| Recruit Holdings | Hold & Observe | One of the top ten contributors in Japan, steady performance | Japanese communications sector |
| Komori Corporation | Hold & Observe | One of the top ten contributors in Japan, steady performance | Japanese industrial sector |
| Toyo Seikan Group Holdings | Hold & Observe | One of the top ten contributors in Japan, steady performance | Japanese materials sector |
| Tanker Sub-Portfolio | Reduce Position | Gradually reduced due to new equity issuance and rising order book | Cumulative return of 440% since 2017, contributed ~80 bps this quarter |
| Refining Sub-Portfolio | Hold & Observe | Greater upside potential, refining cycle still immature | Contributed ~70 bps this quarter, U.S. gasoline demand continues to rise to historical highs |
| Memory Semiconductors | Reduce Position | Cautious on extraordinary profitability, continuing to take profits on price rallies | Specific ticker names not provided |
| Energy | Reduce Position | Oil prices rose 44% this quarter, allowing reduction of positions where pricing already reflects long-term disruptions | Energy exposure at 12% |
| Precious Metals | Rebalance | Using pullbacks to adjust gold and platinum group metals sub-portfolios | Specific ticker names not provided |