AI fears providing new opportunities

During January and throughout the earnings season, data‑centric and information‑rich businesses were caught in an indiscriminate sell‑off as investors fixated on AI disintermediation risk rather than underlying fundamentals. A “sell now and ask questions later” mindset took hold, driving a relentless multiple de‑rating across almost the entire portfolio, despite earnings reports for most Fairlight companies either meeting or exceeding consensus earnings expectations and guidance. During this period of market panic, quality franchises have become a convenient source of funding for rotations into cyclical, capital‑intensive and more speculative areas of the market which Fairlight excludes from its investing process. This has been particularly evident across those businesses that are perceived to be exposed to AI, for example the Fund’s holdings of online classifieds which have been marked down regardless of their actual AI risk profile, degree of technology leadership and product moat. During the month we have leaned into this overreaction, establishing new positions in multiple familiar, quality businesses trading at significant discounts to our assessments of value.

An update on the classifieds thematic

Investor concern continues to centre on the risk that large language models (LLMs) and agentic AI interpose themselves between buyers and sellers, commoditising traffic and eroding the entrenched network effects that underpin these companies’ strong pricing power. Given the high level of uncertainty, quality classifieds assets have increasingly become sources of funding for investors deciding to re-allocate at least some capital towards companies that are seen as benefiting from higher nominal rates, increased spending on armaments and rising commodity prices. The resulting sell-off across online classifieds platforms has been indiscriminate, with valuations compressing to levels that, in our view, embed a severe and undifferentiated AI bear case.

Perceived challenges facing classifieds businesses

Today’s classifieds operators face three principal challenges:

• Risk of market share loss: Agentic AI tools might become traffic gatekeepers, diverting seekers away from incumbent platforms, compressing lead volumes and undermining pricing power.

• Slower macro: In property in particular, lower transaction volumes over the past two years have weighed on agent confidence and marketing budgets. With the global geopolitical scene deteriorating further, transaction volumes are unlikely to recover soon, meaning revenue growth will need to rely more on price increases and new products.

• Several incumbent platforms have underinvested in product and technology in the past, relying excessively on pricing to drive earnings, and are now scrambling to make up for the lost ground. The free cash flow of these companies will likely thus lag earnings growth.

Why AI is a tailwind for vertical platforms, not a threat

We believe many of these fears are overstated, particularly for well-managed, market-leading platforms. It is increasingly clear that LLM providers are viewing vertical platforms as critical partners rather than targets for disruption, given the need for curated, high-quality, and legally compliant data. In practice, two partnership models are emerging: (1) the integration of classifieds platform apps inside LLM front ends, and (2) the use of LLMs to power features on the classifieds platform itself via APIs, with the latter model already live across multiple portals.

Importantly, where platform integrations with ChatGPT have been launched (e.g. recent partnerships Zillow and Scout24), referral traffic from LLMs has remained below 1% of total sessions, and in Scout24’s case, it was just 0.4% in December 2025 (down from its peak of 0.5% a year ago).

The severity of the threat that AI poses varies across classifieds verticals. Generalist platforms are largely insulated from disintermediation risk because the only place their content exists is on the platform itself, while auto and property verticals are structurally better placed than jobs due to the complexity, trust requirements and offline components of the underlying transaction.

Property platforms are currently benefitting from agentic AI by expanding their role along the buying journey - valuation, financing, documentation - rather than being replaced by it. For high-quality operators that are (a) market leaders, (b) well invested in product and technology, and (c) sensibly priced, AI thus looks more like an incremental monetisation opportunity than an existential threat.

Scout24: an AI-enabled ecosystem, not just a website

Scout24, a long-standing holding in the Fund (and one of the largest detractors for the month) is a useful case study of how disciplined, well-capitalised operators can harness AI to deepen their competitive positions.

In 2025, the company delivered 15% revenue growth, 17% operating profits growth, and 20% adjusted EPS growth, while expanding its operating margin to 62.5%. Management expects 16 to 18% revenue growth in 2026 and an organic EBITDA margin of up to 64%, comfortably ahead of the targets laid out at the 2024 Capital Markets Day.

Over the past few years Scout24 has deliberately repositioned itself from a listings portal to a “digital infrastructure player powering German real estate”, underpinned by more than €400m of investment in software, data and connectivity.

At the core of the ecosystem sits Propstack, its proprietary back-end cloud-based CRM software, that is increasingly central to German agents’ workflows, alongside valuation and banking tools used by financial institutions and a Property Hub through which more than 10% of residential units are registered. These systems are embedded in agent day-to-day operations and are not accessible to AI web crawlers, representing a growing body of proprietary data that reinforces Scout24’s information advantage.

On top of this data and software layer, Scout24 has rolled out AI native search and engagement tools across the entire user journey. Semantic and elastic search allow consumers to describe their preferences in natural language, while HeyImmo, a specialised real estate LLM, built on 28 years of behavioural and transaction data, guides property seekers through search, valuation and financing questions in an interactive manner.

In parallel, AI is being used to automate and augment agent workflows: more than 30,000 AI actions have been executed in Propstack, with “voice to listing” reducing listing creation time by over 80%, and the company’s ambition is to evolve Propstack into an agentic AI product that actively prioritises leads, drafts communications and advances transactions.

Margin impact

Far from compressing margins, Scout24’s investment in AI and automation is driving structural operating leverage. Over the past three years, personnel costs have remained broadly flat on an organic basis, despite double-digit revenue growth, as AI tools have effectively given each employee a “second colleague” and shortened development cycles. IT expenses, meanwhile, have grown only in the mid-single digits and represent around 9% of total costs, highlighting the scalability of the platform.

Scout24’s cash generation remains robust, with 2025 free cash flow representing 101% of adjusted net income, and the company has almost no debt.  The board has authorised a €500m share buyback programme (close to 10% of the current market cap), underlining management’s conviction that the intrinsic value of the business exceeds the current market valuation. Management and directors have also been buying shares with their own money over the past few months.

The Fairlight View

In January we highlighted that almost all portfolio holdings now trade at or near the trough multiples of the past decade. At the Fund level, our P/E multiple has compressed from 24x to 19x, reducing the historical valuation premium to the benchmark from eight turns to just one, despite superior operating metrics and lower leverage.

The current sell-off has provided an opportunity to add to high-quality franchises at valuations that offer attractive future returns and further upgrade the quality of the portfolio. The combination of indiscriminate selling, sector-wide multiple compression and resilient earnings growth has produced, in our assessment, a favourable asymmetry: downside is increasingly underpinned by cash generation and buybacks, while upside is linked to a normalisation of sentiment and continued EPS compounding.

Scout24 exemplifies this dynamic. The company is gaining share, expanding margins, and deepening its competitive moat through AI-enabled products and proprietary data, yet trades on a valuation that gives it little credit for these characteristics.

In this context, we are reminded of Warren Buffett’s observation that “The stock market is designed to transfer money from the Active to the Patient”. The market’s current impatience - expressed through broad-brush selling of anything perceived to carry AI risk - stands in contrast to the discipline of our portfolio companies, which are investing in their competitive moats and returning capital to shareholders where appropriate.

We intend to remain on the side of patience, using volatility to increase exposure to businesses that we believe can continue to compound intrinsic value at double-digit rates, including select classifieds leaders such as Scout24.

As we look ahead, our focus remains on three things: (1) owning companies with demonstrable pricing power and strong balance sheets; (2) ensuring our holdings are on the right side of technological change, including AI; and (3) maintaining valuation discipline. We believe this approach will allow the Fund to benefit as the market eventually shifts from “voting” on AI narratives to “weighing” the cash flows of individual businesses.