2025 Year-End Letter: Quality Strategy

With another year of market ebullience behind us, January seems a good time to take stock and share our thoughts on the portfolio for the years ahead.

Our flagship GMO Quality Strategy delivered 19.4% net of fees from a U.S. dollar perspective, ahead of the S&P 500 but behind the MSCI World index. The U.S.-only and international variants of the strategy generated broadly similar absolute returns.

It will surprise you very little to learn that the best returns were generated in the growth part of the portfolio. The top three performers were AI-related. Lam Research and KLA Corp both sell equipment essential to the manufacture of semiconductors necessary for all that AI compute. Their businesses accelerated meaningfully in the scramble to build out new data centers.

Alphabet, meanwhile, had a great year on multiple fronts: its TPU chips emerged as a potential competitor to Nvidia’s GPUs; its Gemini AI product took the lead over OpenAI’s best offering in benchmarking tests; and the widening of the search market into the AI space took regulatory heat off its powerful position in conventional search.

AI

Overall, AI made a strong contribution to portfolio returns. While there is no perfect measure of AI exposure for a portfolio, our best estimate is that the portfolio has been a couple of points underweight over the year relative to the S&P 500 and a little overweight vs. the MSCI World.

However, the relative weight mattered less than owning the right stocks in the space. Given that AI is likely to remain a key issue for some time, we thought it might be helpful to describe an exercise we undertook last year to reassess AI’s impact on company quality amongst those most exposed.

We classified the AI ecosystem across four layers: applications, LLMs, compute, and data center suppliers, and used this to build a model of how money flows, and might flow in the future, through the corporate plumbing of the industry. Each layer’s capex and opex drives the revenue of the layer below it.

This approach enabled us to model the implications of various scenarios and to develop principles for assessing the strengths and weaknesses of the different industry participants. On balance, the most useful considerations for us were:

• concentration of revenue by source (AI vs. non-AI) and by customer,
  
  
• capex requirements vs. funding ability, and
  
  
• revenue exposure to capex contingent on AI growth, as opposed to more stable maintenance or “keep the lights on” capex.

These proved helpful in strengthening our assessments, particularly in the compute and data center supplier layers, reinforcing our sense that the hyperscalers have sufficient flexibility and breadth to handle the uncertainty as to how the area evolves.

Within the data center supply layer, we retain a preference for companies with deeper non-AI businesses and more diversified AI exposures, for example, preferring TSMC to Nvidia. Most significantly, the use of this set of lenses led to the liquidation of the portfolio’s Oracle holding in October.

We had held Oracle’s stock continuously for 18 years from 2007, often as a top-10 position. For most of the last decade, we viewed Oracle as priced for a respectable but unremarkable future, but with the possibility of faster revenue growth stemming from the cloud transition in enterprise software. That growth proved elusive for many years, and the stock was priced on a low teens multiple as recently as 2022.

The slow revenue growth didn’t bother us unduly, as the company generated a high return on capital and thus delivered earnings growth in most years through margin improvement and share buybacks, alongside a rising dividend. Meanwhile, Oracle stayed broadly relevant via both inorganic and organic investments at a meaningful scale.

However, over the last couple of years, Oracle’s prospects were transformed by demand for its data centers. Oracle arrived later to the cloud party than Amazon and Microsoft, but it had a more modern cloud architecture as a result. When AI-related demand for compute emerged in 2023, the company’s reputation for efficient data processing, strong security, and its large installed base of Oracle’s databases created an opportunity for explosive growth. Management grasped it with both arms, and AI investment began to dominate shareholder considerations.

The capital requirements on the business continued to intensify (capex in 2026 is expected to be 10 times higher than it was in 2022!), and leverage grew, gradually in the form of conventional debt and exponentially in terms of leases and off-balance sheet financing. In September, when Oracle revealed the extent to which their compute business was dominated by a single client—OpenAI—we downgraded our assessment of the company’s quality to below our minimum threshold and liquidated the position over the next month or so. Oracle’s stock fell 43% from its October high to the December low, validating our rationale so far.

Turning back to our thought experiments, for the applications layer, the levers are harder to identify, and many of the winners are presumably yet to emerge. Picking start-ups is beyond our remit, but we do need to assess the ability of existing software companies to compete, or perhaps more importantly, deter potential new entrants.

Here we adopted a second approach, reasoning that the proximity of a software company’s product to critical corporate data (think: systems of record) will likely become a key determinant of AI success in the years ahead. Why? To use AI effectively, corporations need to get their data in the cloud and in the right format.

This is no small task. We expect that: a) companies are unlikely to want to add further complexity by switching software providers at the same time, and b) the time scale involved (years) means that the incumbents will have plenty of time to perfect their agentic offerings.

This conclusion is interesting as it is contrary to current market opinion. In 2025, enterprise software businesses of this type performed poorly. The weakest three technology stocks in the portfolio were in this category: Salesforce, Dassault Systèmes (the French lynchpin of industrial engineering design software), and Accenture (not an enterprise software provider per se, but whose corporate IT work is closely related). Each struggled for slightly different reasons, but these stocks, and others like them, have derated as a group.

Market participants worry about the risks from AI-native new solutions in enterprise software, presumably vibe-coded, and perhaps simply see more exciting ideas with stronger short-term tailwinds elsewhere. The incorporation of agentic AI into enterprise work will evolve over the next several years, and we suspect that the enterprise software businesses have time to become agentic suppliers of choice.

With valuations running well below recent levels, we don’t have to be precisely right. We smell an opportunity emerging here and have recently added capital to positions in enterprise software, including one new stock for the portfolio. ¹ We watch with interest.

Healthcare

Whilst returns in the Quality Growth part of the portfolio were stellar in aggregate, the results in the other silos (Quality Value and Core Quality) were generally more grounded. There were exceptions.

In the value bucket, Safran, the dominant manufacturer of narrow-bodied jet engines via its joint venture with GE, continued to perform well, benefiting from strong aftermarket trends and improving profitability of its new engine model. In the core bucket, Apple rallied strongly from April lows as the tariff haze cleared. However, it is in one of the areas that disappointed in 2025 that we see some of the most promising sources of future returns.

Healthcare spending as a share of GDP has been rising materially for at least 50 years—not just in the U.S., but across the OECD countries for which we have good data. It might be convenient to explain that away as a rapacious industry squeezing its consumers, but the reality is less sensational, though warmer and fuzzier. Healthcare spending has risen while the prices of many consumption items have stagnated because society places a high value on healthcare, and the demand for innovative medical solutions is very real.

That may sound like a good setup, and it is, but investing in healthcare can be a messy affair with political, scientific, operational, epidemiological, and other issues frequently obscuring the bigger picture. 2025 was tough for the strategy’s healthcare positions in aggregate. We believe that the healthcare sector represents a rich seam of opportunity at this point, across a variety of different areas, but the old market saw, “you can’t have good news and good prices at the same time,” comes to mind.

Managed care companies had a number of problems (some external, some self-inflicted) and were the most significant pocket of weakness. We previously wrote about the challenges for this U.S.-centric sector, which sits between the consumers of healthcare (insured plan members) and the payers (usually employers or public entities). ²

In this year of political maneuvering, the ever-present discount widened further. To cap it all, UnitedHealth delivered sufficiently disappointing results to prompt a change in CEO mid-year and halve the share price to boot. We believe they have embarked on a course of self-help that will ultimately prevail and took the opportunity to acquire more UNH stock.

Life sciences equipment makers have been on the sharp end of an “R&D winter.” The post-COVID biopharma boom turned into a bust of once-promising treatments and technologies. Further damaging sentiment was a series of laws and regulatory changes from both Biden and Trump: biopharma companies feared price negotiations and tariffs, while universities struggled with NIH budget cuts and a broadly difficult financial environment. But there are signs of life as the regulatory environment stabilized, and the strategy’s holding in Thermo Fisher was one of the strongest performers in the second half.

Pharmaceutical companies also had a mixed year and found themselves caught in the presidential crossfire over issues such as non-U.S. manufacturing for U.S.-destined products and the relative pricing of medicines in different markets. By the end of the summer, the “Pfizer playbook” had emerged, which includes promises to relocate manufacturing, discounting of drugs in the U.S. for certain channels, and addressing of pricing disparities across borders. This took away most of the heat and, on average, the strategy’s pharma holdings had a good year.

The strategy has a substantial allocation to healthcare, across various sub-segments and investment types, from compounders in Core Quality like Johnson & Johnson to the high-growth, undisputed leader in robotic surgery, Intuitive Surgical. We believe that investment in medical innovation will tend to be rewarded, especially at this point when the benefits of that innovation are befogged by other issues. Healthcare stocks began to pick up in the fourth quarter, and we hope that is a taste of things to come for these well-positioned companies.

Navigating Choppy Waters

Our investment approach is built around two key tools that serve as a lodestone of sorts. Firstly, we emphasize company quality as a way to access the fundamental returns associated with disciplined capital allocation and competitive advantage. Secondly, we keep valuation front of mind to reduce the risk of overpaying for that quality.

We do not attempt to know everything about everything. We limit our investments to companies that we deem to be sufficiently high quality. That rules out whole sectors, markets, and swathes of companies, whatever their valuation. By doing so, we hope to bias the portfolio toward strong businesses while keeping the number of things to think about manageable for the team.

With our relatively concentrated portfolio of quality names, we try to focus on the issues that matter on a multi-year basis, i.e., those of significant importance to the company in question. This puts us in a different posture than those who try to make hay from better forecasts of quarterly earnings. Company results, therefore, serve us more as an update on a multi-year process of capital deployment and competitive advantage than a signal as to where to move capital today.

This approach has helped us to navigate the, at times, overwhelming array of news flow in 2025. We tend to fade (trade against the prevailing trend) the majority of market-jerking headlines, as most things matter less than the market hopes or fears they will.

During the tariff-related sell-off, AI data center suppliers were hit hard for their reliance on Asian manufacturing. The selling had an indiscriminate feel to it, and we wondered to ourselves whether investors would be taking tariffs so seriously in a few years’ time. We took the opportunity, for example, to add to Broadcom, a substantial importer of tariffable semiconductors and related goods into the U.S.

The trade proved to be well timed in retrospect, but we had no sense that this would be such a short buying window; rather, that if the market was in a flap, it would likely be a good idea to flip, in this case, from some of the strategy’s most conservative positions into more constructive ones. If the sell-off had continued, we would have traded against it harder.

It would be foolish to dismiss all developments—the future is the cumulation of yesterday’s news, after all. The AI transformation is real. Healthcare innovation is changing the world patent by patent, or patient by patient. Even then, one must maintain a detached view.

At the time of writing, DRAM and NAND stocks were being pushed ever higher by the salivating hordes. AI’s need for high bandwidth memory, for example, has diverted manufacturing capacity away from the less exotic memory chips, and as a consequence, the price at the lower end has been propelled skywards. Is that a multi-year effect? Or will capital be attracted to “correct” the prices of these modern commodity goods?

We expect the strong businesses in the portfolio to continue to generate returns on capital going forward. No doubt their share prices will lurch this way and that as they do, and we will seek to rebalance the portfolio to the areas where the combination of valuation and company quality is most compelling, endeavoring to understand what is ephemeral and what is significant.

Despite the tariff trauma, AI acceleration, and all the other crosscurrents, the Quality Strategy got the return it deserved in aggregate in 2025; the return achieved by the strategy closely matched the fundamental return generated by the companies in the portfolio (as measured by growth in EPS and dividends received over the period). Because trading was additive to returns—adding to Broadcom and liquidating Oracle and so on—portfolio valuation perhaps got a little more attractive over the period.

We believe that the strategy is well-positioned for the quarters and years ahead. Despite all the uncertainties, if we can build a portfolio of companies deploying capital effectively, with the reassurance that their valuations are under watchful review, we hope that the GMO Quality Strategy can continue to play a helpful role in client portfolios.

Download here >