Investors are still underestimating the long-term impact of AI

The writer is head of thematic research in Europe at Morgan Stanley

A meme now circulates on social media: Atlas holding up the globe. But instead of earth it is the entire stock market and instead of Atlas, it is the theme or stock of the moment, currently artificial intelligence. For the bulls, it is a victory lap, and a hope of more. For the bears, a warning of the current levels of market concentration in a handful of stocks and thematic froth.

The shares in companies that enable AI, such as chip producers, have been stellar performers over the past 18 months. Yet, echoes of the dotcom era continue to be debated ad nauseam.

To get bogged down in this debate is to misunderstand the dependability of Amara’s Law. Named after an adage of the futurist Roy Amara, this states that we tend to overestimate the impact of technologies in the short run and underestimate them in the long run.

Financial analysts routinely underestimate adoption of general-purpose technologies — of which AI is undoubtedly one. Our research shows initial forecasts of the market sizes of internet, the cloud, smartphones or PCs after 10 years were wrong by around 40 per cent — with remarkable consistency. Our view is that AI will be no different, except that estimates are likely wider of the mark than any of these forerunners.

We have believed for the best part of a decade that data technologies, such as AI, would diffuse across all economies, unlocking productivity in almost every industry. This is now happening. The cost to query generative AI models has fallen 90 per cent since 2021. This deflation accelerates adoption and has ripple effects across the economy

Yet, the Achilles heel for AI bulls is the same as the one fuelling AI bears: there is no simple way to track AI’s adoption nor the success, or otherwise, of its implementation.

The market is unable or unwilling to price any adoption gains ahead of delivery. Consequently, companies adopting AI have underperformed the broader market. Over the long term though, it is implausible that many, if not most, of these adopter companies would not benefit from a productivity upswing.

But even if we lack a catch-all metric to track adoption or failure rates of projects, of which we appreciate the number will be high, examples of AI adoption are beginning to come in thick and fast.

L’Oréal recently reported how an AI-powered app delivered a 63 percentage point uplift in the conversion rate of inquiries into sales compared with traditional counter selling methods. Launching commercially in April, the app makes personalised beauty product recommendations to the user.

A leading coding program that uses AI has been saving research and development departments up to 55 per cent of their time. A prominent fintech saw an 82 per cent improvement in customer service resolution speed in the first month of using an AI programme. The list goes on.

With cheap generative AI tools, for example, companies can perform live translation with eye contact and lip synching to give seamless presentations in almost any language. Imagine the implication for the addressable market and customer stickiness of video streaming platforms. A logistics company even put this to use last month, translating their earnings release into five languages simultaneously. A gimmick to some, perhaps, but game-changing for a company catering to clients in 174 countries. The inclusionary benefits of this technology are not lauded enough.

For now, the market is still treating AI enabler companies as the only game in town. We have not seen the last of the Atlas memes. But short-term scepticism around enablers’ valuations in the early innings of secular themes risks missing the wood for the trees. Since 1990, missing the 30 best days of the S&P 500 cuts your compound returns in half. Whether via the household name enablers or the yet-to-be-appreciated AI adopters, the best long-term strategy is to stay invested.

Our view at Morgan Stanley is that the combination of accelerating AI technology development and the need to address widespread productivity challenges means that AI is one of the most important investment themes of this decade. We see 2024 as an investment year that will be followed by measurable productivity improvements beginning in 2025.

Markets will, of course, judge themes swiftly and harshly when they slip up. That is not unique to AI. For longer-term investors, though, this theme is best encapsulated by the saying: “Pessimists sound smart; optimists make money.”