Please refer to our disclaimers, which can be found in the footnote of this page and here.
2Q 2025 performance was 8.7% gross / 8.4% net (vs. the MSCI ACWI at 11.5%). Since inception (1 Jan 2019), the portfolio has compounded at +18.8% gross / +15.9% net (vs. the MSCI ACWI at +13.3%), representing +5.5% gross (+2.6% net) annualised outperformance. The top contributors year-to-date were Alibaba, Microsoft, Meta Platforms, and Visa. The top detractors were Salesforce and Alphabet.
We started the quarter with 13% cash. As noted in our prior letter, during the early April sell-off, we added to Salesforce, Alphabet, Meta, Microsoft, and Block. Cash dropped to 5.7% of the portfolio. Post the market recovery, we exited our remaining position in Snowflake (2% of portfolio), and we trimmed Alphabet (17% to 10%) and Meta Platforms (7% to 5%). Our sizing decisions are influenced by 3 factors — business quality, valuation, and conviction (discussed in our Investment Process overview). The Snowflake and Meta Platforms decisions were purely valuation-based, while the Alphabet decision was driven by reduced conviction given the emerging AI threat to Google Search. Following these sales, cash stands at 18%. This is an uncomfortably high level of cash since it is a poor economic asset over the long-term. However, we are close on a few new names and expect this to come down.
Our Snowflake investment journey is shown below. We had followed this position for a long time before owning it (while still a private company). We have owned it twice since the inception of the strategy. We made the first purchase in early 2021 but sold later that year, for an IRR of 32%, as technology growth stocks entered extreme bubble territory. We re-entered Snowflake after the bubble burst in 2022 though we were too early. We made several additions since then as the stock price experienced very significant volatility. The IRR the second time around was only 11%. The IRR for our day 1 clients was 33%.
We have held Alphabet since the inception of our strategy — and indeed, for many years prior. This commentary is not intended as a comprehensive review of the business, but rather to provide context for why we reduced our position in Alphabet after such a long holding period.
Google Search, Alphabet’s core business, is a toll road for the internet. Its competitive position has been entrenched via positive network effects whereby billions of Google Search users provide important feedback loops to improve the search algorithm. The ability to reach billions of users with rich intent data creates strong advertiser demand and provides Google with the cash flow to reinvest into its value proposition. Google used its privileged position to reinforce its dominance through means that some consider anti-competitive (as discussed in our note on Google’s Antitrust Battles). This includes building, acquiring, or renting key parts of the user and advertiser funnels — resulting in a global search market share of ~90% (80% desktop, 95% mobile).
Simplified Search Funnel (LHS) and Digital Advertising Value Chain (RHS) Source: GS Statcounter, Datanyze, NetMarketShare, DOJ, Apptopia, Kalakau Avenue Estimates Note: Less relevant elements removed for simplicity
The rise of generative AI represents a credible threat to Google Search’s dominance. Generative AI enables users to bypass the traditional search paradigm altogether — engaging with the internet in a far more intuitive way. Crucially, while Alphabet is a leader in AI, it does not command the kind of monopoly in this space that it enjoys in traditional search.
Google has weathered platform transitions before — most notably the shift from desktop to mobile. Google recognised this transition very early on and managed to gain control of upstream mobile search funnels. However, it is often overlooked that Google’s actual share of total user search activity fell during that period (see diagram above). The high market share levels mentioned above reflect browser-based searches only and exclude direct search activity on retail platforms, travel apps, social media, etc and increasingly, generative AI interfaces such as ChatGPT and Gemini. Despite losing total share during the mobile transition, the overall growth in search and digital advertising was more than enough to offset this, and Google Search revenue continued growing strongly. Given the greater maturity of these markets today, loss of share to generative AI search will have a more noticeable impact. Further, unlike the mobile transition, where Google’s core engine remained the centre of gravity, generative AI search represents a break from that model.
Google is responding to this threat in several key ways. First, it introduced AI overviews into its search results, which has now reached 1.5b monthly users, with monetization at approximately the same rate as traditional search. Second, Google is rolling out AI Mode, an AI chatbot interface accessible from the Google home page. Third, Google has launched Gemini, its direct competitor to ChatGPT (OpenAI), which now has 400m monthly users and an estimated 45m daily users. By comparison, ChatGPT is estimated to have surpassed 1 billion weekly users (this would be even higher on a monthly basis). ChatGPT receives 5+ billion browser-based site visits per month versus Gemini at 650m (though outgrowing ChatGPT). Google also benefits from generative AI demand via its cloud infrastructure and AI models. Despite playing catchup to OpenAI, there are positive signs in all of these.
That said, Google continues to prioritize traditional search. We see AI Overviews as an incremental improvement — not a transformational shift. True generative search experiences (e.g. ChatGPT, Gemini, Perplexity etc) are likely to offer more compelling user value in the long term. The big question is whether traditional search, supplemented with AI overviews, can continue to grow while ceding share to true generative AI search.