AI, platforms and the battle for the customer
For some time, investors have pondered the AI revolution in terms of relative winners and losers, who will spend the most to “win” and whether “winning” presents a Pyrrhic victory in the sense that to win the AI arms race - if you are, for example, a cloud infrastructure provider - companies may risk overcapitalising, with a subsequent collapse in returns on capital. These are important questions, but a more recent dynamic has emerged too that warrants attention.
AI is not just another productivity tool. It has the potential to change how customers make decisions, and in doing so, quietly rewrite the economics of some of the world’s most valuable platform businesses.
For the last two decades, tech platforms have thrived by occupying the space between intent and action. Whether booking travel, shopping online or searching for information, value was created by owning that consumer journey. Control the search, comparison and transaction, and you control pricing power, data and, ultimately, profit pools.
Recently, some understandable concerns have emerged that AI tools such as Claude threaten this model not by replacing platforms outright, but by abstracting them away. So, the thinking goes: as AI-powered agents become more capable, the customer no longer needs to navigate multiple websites, compare dozens of options or even engage directly with a platform’s interface. The agent does that work on their behalf. The decision is compressed into a recommendation, often with little visibility into the underlying marketplace. This matters because the platform’s advantage has never been the product itself, but the friction that it removed. If AI removes that friction instead, the platform’s role becomes less visible and potentially less valuable.
From an investment perspective, this forces a more nuanced assessment of platform businesses.
Take Booking Holdings as an example. At first glance, travel bookings might appear vulnerable to total AI-led disintermediation. But Booking’s strength does not rest solely on directing traffic. It also has the task of building and maintaining supplier relationships, managing inventory in real time and – perhaps most importantly – acting as a merchant of record. The latter is critical when travel plans get disrupted (as is no doubt currently happening) – the consumer will want peace of mind about who they have contracted with and that alternative arrangements can be made, or refunds issued (as in COVID). This is no simple matter to replicate – after all, Google tried for years and ended up being happy to earn a slice of the economics via performance marketing for Booking. In this context, AI may well alter how customers arrive at a decision (i.e. the discovery process), but it does not eliminate the value of the underlying marketplace. What may well happen is a change in who receives marketing revenues if traffic is, for example, diverted away from traditional search towards an AI chatbot. We’re pretty certain Anthropic and OpenAI wouldn’t mind receiving a slice of the economics of travel engines if they become responsible for much of their traffic via performance marketing fees; conversely, we are much less sure they would want the complexity of managing the OTA business model from end-to-end. The end result may or may not result in incremental margin pressure for companies such as Booking, but the likelihood of an extinction-level event playing out is low in our view.
These questions sit at the heart of recent volatility across technology and platform stocks. Markets are not simply reacting to earnings revisions or short-term news flow – in most cases, we’ve seen no evidence of recent “disruption” showing up in results. Rather, investors are making judgement calls about the terminal value risk of businesses that have long held strong moats, but where the confidence in those moats has been shaken given question marks about where economic power may land as AI reshapes customer behaviour. When long-held assumptions are challenged, valuation ranges tend to widen before they narrow.
Periods of structural change also amplify investor behaviour. Early in the cycle, optimism is often rewarded - capital flows towards the most visible innovation, the boldest spending plans and the “cleanest” narratives. As uncertainty rises, that enthusiasm can reverse quickly, leading to sharp corrections and, at times, indiscriminate selling across entire sectors - which is exactly what we’ve witnessed recently across the broad software and tech platform space.
Volatility in this context is not evidence that opportunity has disappeared, but that it is being re-priced as expectations adjust.
It also reinforces an enduring investment principle: technological change does not remove the discipline of valuation - it heightens its importance. Businesses that combine innovation with pricing power, strong cash generation and the ability to reinvest at attractive rates tend to emerge with their economics intact, even if their role in the customer journey evolves.
Exposure to long-term innovation remains important, but it must be balanced with an understanding of business quality, competitive advantage and downside risk. As ever, durable outcomes are not a function of correctly predicting the future – nobody we know can do that. Rather, they are about making informed judgement calls about the extent to which expectation resets have been captured in valuation, combined with managing risk and maintaining exposure to diversified sources of value creation at a portfolio level.
