News & Insight • Market Insights
16 October 2025 | William Buckhurst
Thoughts on Quarter 3 2025
YFS VP Portfolio Funds
Over the last three months, the Vermeer Partners Portfolio Fund returned +4.73% and the Growth Portfolio Fund (which will have a higher equity content and, therefore, higher risk profile) returned +5.84%. The estimates for the ARC GBP Steady Growth PCI were +5.07% and the ARC GBP Equity Risk PCI 6.20%.
Since launch (16th October 2023), the Portfolio Fund has returned +15.70% while the Growth Portfolio Fund has returned +15.22%. The estimates for the ARC GBP Steady Growth PCI were +21.78% and the ARC GBP Equity Risk PCI +25.22%.
The estimated historic dividend yield on the Portfolio Fund is 2.91% and on the Growth Portfolio Fund it is 2.30%.
Portfolio Review
“they strained their chests against enormous weights, and with mad howls rolled them at one another. Then in haste they rolled them back, one party shouting out:
“Why do you hoard?” and the other: “Why do you waste?”
So back around that circle they puff and blow.”
Dante’s Inferno, Canto VII, The Divine Comedy (transl.)
In his 14th-century epic poem, The Divine Comedy, Dante Alighieri structures Hell through nine descending circles, with the outermost punishing lesser sins and the innermost reserved for the most grievous transgressions. While Dante's vision of eternal damnation may seem extreme – and medieval Italian literature rarely features in our daily reading – it offers a striking allegory for today's markets, where circular capital flows have raised serious questions about the sustainability of the AI and broader technology rally.
We wrote in the first quarter about the sharp reversal of this trend Q1 Review, then witnessed its reignition in the second quarter Q2 Review. Now, nine months into the year and with year-end in sight, we believe it is crucial to examine what many perceive as a classic bubble forming. The stakes are high: last quarter, Nvidia, Broadcom, Tesla, and Apple alone contributed over a third of the 7.5% return in global equity indices. Not owning just these four stocks leaves the rest of any portfolio facing an uphill battle to beat the index – nonetheless we were pleased with the Q3 performance which was only slight behind the (estimated) ARC indices.
In 2024, Nvidia invested approximately $1bn in AI startups globally through direct investments and its venture capital arm, NVentures – startups that predominantly buy Nvidia chips. This was just the beginning: hyperscaler spending has surged from roughly $125bn two years ago to around $200bn in 2024, with consensus expectations exceeding $300bn in 2025. This evokes memories of a classic "vendor-financing circle," where capital raised by companies is recycled back into the same AI ecosystem, raising questions about whether Nvidia and others are investing heavily to prop up demand and sustain their own markets, creating a loop where the same dollars circulate rather than representing genuine, independent customer demand.
The most notable example is Nvidia's recent partnership with OpenAI, in which Nvidia can invest up to $100bn over time. The circularity is stark: OpenAI purchases Nvidia hardware, Nvidia reinvests those profits back into OpenAI, which then uses those funds to buy even more Nvidia equipment.
Nvidia's partnership with CoreWeave reveals an even more intricate web. CoreWeave secured debt financing using Nvidia GPUs as collateral. Nvidia owns shares in CoreWeave. Nvidia also struck a deal to use any of CoreWeave's excess capacity through to 2032. Again, Nvidia's investments enable CoreWeave to expand its data centre operations further – which means purchasing more GPUs from Nvidia.
We could go on: Oracle's deal with OpenAI, IBM's partnership with Anthropic, OpenAI's agreement with Advanced Micro Devices, and so on. Never before has so much capital been deployed so rapidly on a technology that, for all its incredible potential, remains largely unproven as to how it can generate enough return on investment.
We do take some comfort, however, that compared to previous boom-and-bust cycles, much of this capex derives from free cash flow. So far, only Oracle among the major publicly traded technology companies has resorted to debt financing for its massive data centre buildout to service AI cloud clients. This represents an important distinction from the uncomfortable parallels with the late-1990s telecom bubble, where enormous capex on fibre optics – much of which has never been activated – was financed through debt or dilutive equity raises.
The other critical difference is asset longevity. Unlike previous cycles, the products being invested in today have remarkably short lifespans: GPUs have a useful life for frontier applications such as model training of around three years. Roughly a third of hyperscaler capex is into such short-lived assets, which remain – in theory – monetizable afterwards. These shorter depreciation cycles should impose financial discipline on investors far more quickly than in past manias, but we have no evidence it will. During the railway boom and bust of the late 19th century, asset lives spanning decades or even across centuries (US freight was still being run over 19th century track well into the 20th century) masked the weakness of many business models, allowing railroad companies to stumble on for years before insolvency. In AI, the tide could go out far sooner.
Former General Electric CEO Jack Welch famously posited that only two players can achieve profitability in a truly competitive industry. Below the top two, survival becomes a struggle. According to this argument, for which we have some sympathy, at least three and perhaps more of the current LLM players will ultimately be forced to write off their investments.
We accept that we have been underinvested in the broader technology and AI theme over the past two years, and this has been a major reason why, despite acceptable absolute returns, our relative performance has been underwhelming. Perhaps we have been overly mindful of past bubbles. Rarely, if ever, has insatiable demand growth occurred without being accompanied by over-investment followed by excess-capacity leading to sharp share price declines. Microsoft shares peaked in March 2000 and took 16 years to recover.
Nonetheless, mindful of AI's importance and long-term growth potential, as well as the financial robustness of many companies in the area, we have maintained a meaningful sector weighting. We own Microsoft, Alphabet, Amazon, ASML, and KLA Corporation, as well as electricity infrastructure plays such as Siemens and Schneider Electric – the first three possess incredibly robust balance sheets, while the latter four are equipment manufacturers or "enablers" that – at the current rate – benefit from the enormous capex cycle. We also hold long-standing investments in technology-focused funds such as Scottish Mortgage and Schiehallion, while some of our Asian and Emerging Market funds maintain meaningful positions in technology winners like TSMC and Tencent.
However, by any measure, we remain underweight the theme but, given the risks highlighted above, are comfortable with that stance going forward. Indeed, even as AI-related stocks powered ahead last quarter, we reduced our large Microsoft position and exited IBM entirely.
Meanwhile, we are already witnessing what could be the early casualties of AI disruption – what one of our colleagues has touchingly termed "AI roadkill." Adobe, developer of Photoshop and Acrobat, faces serious threats from generative AI and emerging graphic design platforms such as Figma and Canva. Salesforce and other Software-as-a-Service (SaaS) providers are similarly vulnerable. Yet potential opportunity lies within this disruption. Many displaced companies will present chances to invest in solid businesses at depressed valuations – they are not obsolete, merely suffering from the same exaggerated share price movements driving so-called "AI winners" higher. Many of these tools are deeply embedded in business practices and could prove difficult to displace, even if AI upstarts offer cheaper alternatives.
Similar unanswered questions surround owners of market or professional data services such as MSCI, London Stock Exchange, and S&P Global. We continue to devote significant time and resources to understanding whether AI represents a threat or opportunity in these areas. As an example, following a share price fall we have made a new investment in RELX whose dominant legal database LexisNexis appears to be under threat. But while AI can generate in-depth corporate research reports, the legal industry requires tools that attribute sources, verify facts, and record workflows, which LexisNexis provides and will be difficult for AI to replicate. We are similarly confident that an investment we made in Accenture at the start of the year will prove successful: despite concerns over AI disruption (as well as DOGE-related cuts to US government departments), its last results continue to show impressive revenue and EPS growth.
We run balanced, multi-asset portfolios and it is important not to become too focused on one area (at a quick count AI, has already been mentioned 18 times in this relatively short thought-piece!). There has been quite a lot going on in Japan too – at Vermeer we have held a long-standing overweight in Japanese equities. The corporate reform story is now well rehearsed – government led policies to streamline Japanese corporates, unwind cross-shareholdings, boost profitability and unlock shareholder value have succeeded to the extent that, in many cases, this has become voluntary rather than mandated. Japan has been home to many fantastic businesses that were trading on cheap valuations relative to international peers and are now beginning to show a real focus on shareholder returns. We do not believe this phenomenon has played out and still see good opportunity for growth in our two Japanese funds, and three direct securities. And not least, our Japanese sleeve provides a good counterbalance to our more fashionable AI-realted names. There, we’ve mentioned it for a 19th time.
Portfolio activity
We made some modest changes to your Funds over the quarter. New additions included Eli Lilly, AJ Gallagher, Informa and RELX.
- Eli Lilly is now the clear leader in Obesity and Diabetes. Mounjaro, its type 2 diabetes drug, is branded as Zepbound for obesity. It’s a GLP-1/GIP receptor agonist, part of the new wave of highly effective weight-loss drugs competing directly with Novo Nordisk’s Ozempic and Wegovy. Sales growth has been explosive, addressing massive global markets: it is estimated that c. 650 million people are classified as obese worldwide. A pull-back in the share price offered a good entry point in July.
- AJ Gallagher (AJG) has grown to become one of the largest insurance brokers in the world, competing with Aon to be number two behind Marsh. A slightly softer insurance market in recent months (property renewal premiums have declined whereas other areas such as casualty, auto and general liability continue to see firmer pricing), has led to a de-rating in the shares – they trade on a forward p/e ratio of 23x compared to a recent average of around 30x. This presented a good buying opportunity for a market leader with significant pricing power that should continue to grow revenues at low double digits year in year out. It also has displayed incredibly impressive capital allocation decisions with successful bolt-on acquisitions over many years, operating in a market that continues to be ripe for further acquisitions.
- Informa is a UK-listed but multi-national company (95% of revenues are overseas) with a market cap of over £11bn, operating across B2B live events, academic publishing, digital information services, and specialist market intelligence. At the time of purchase the shares traded on 15x forward earnings (which is cheap for the current environment) and were on a free cash flow yield of 7.2%.
- We sold the holdings in Halma, Ashtead Group, and IBM. Halma and Ashtead are both very good businesses but following strong rallies in their share prices we felt the valuations were up with events and so have taken profits but would revisit at a more attractive price. Aside from its historic hardware division, IBM has positioned itself as an infrastructure provider for enterprise AI rather than just a pure cloud play. But it trades on a higher valuation now for what are unremarkable growth metrics relative to other technology stocks.
- We also switched the holding in BB Biotech into the BlueBox Precision Medicine Fund. The BlueBox Precision Medicine Fund is a relatively new $38m actively managed fund investing predominantly in sub $50bn market-cap biotech companies that develop targeted (“precision”) drugs – a fast-growing area of healthcare that uses a patient’s genetic, environmental, and lifestyle data to tailor personalised treatments and diagnostics.
- On the Portfolio Fund, which invests for a higher income yield, we added a new holding in Bluefield Solar Income (BSIF) which is a c. £600m investment company that owns and operates utility-scale UK solar assets with around 883 megawatts (MW) of renewable power and 27 years remaining asset life. On a sunny day, it estimates that it is responsible for around 4% of the UK’s electricity demand. It also has around 10% of its portfolio in wind assets. The shares trade on a c. 20% discount to NAV, and the merchant power price has been weak of late (albeit from very elevated levels in 2022), so we feel that this is a particularly opportune time to invest. The dividend yield is round 10% and it has grown its dividend every year since launch in 2013. As an alternative, this should provide an IRR healthily in excess of the 10-year UK gilt and / or cash.
- Finally, as part of our strategy to spread risk across different themes, sectors and asset classes, we have had a long-held position in a gold ETC across both Funds. For years it did very little (the insurance premium that didn’t pay out - despite Brexit, Trump 1.0, covid, wars and inflation). Well now it has. The gold price has doubled since the start of 2024. Perhaps all it took was a President who launched a trade war against most of the world as well as appearing to want to attack the independence of the Federal Reserve. Regardless of what drives the gold price higher or lower, we are thankful to have some.