AI Is No Longer Science Fiction: The Investment Opportunity Nobody Wants to Miss

AI has quietly left the laboratory. The harder question for investors is what exactly they are buying when they buy “AI”.
This was supposed to be the year everything stopped. Instead, it became the year in which artificial intelligence stopped being a research topic and started being a line item in capital allocation decisions.
The numbers speak clearly enough. Preliminary estimates suggest that AI startups attracted around $74 billion of venture capital in 2020, roughly one in every five dollars invested by VC funds globally. The United States and China account for the overwhelming majority of that. Europe, despite its research base, remains a distant third.
Behind the money, the technology has also changed character. In June, OpenAI opened access to GPT-3, a language model with 175 billion parameters that can write passable text, summarise documents and even generate simple code. A few weeks ago DeepMind’s AlphaFold produced results in the protein structure prediction competition CASP14 that many biologists did not expect to see for another decade. These are not product announcements. They are signals that the frontier is moving faster than most business plans assume.
Three kinds of AI investment
When I look at the deals being done this year, I see three quite different things hiding under the same label.
The first is AI as a product. Companies whose core offer is a model or a platform: computer vision, speech, language, prediction engines. This is where the most visible valuations sit, and where the competitive pressure from the large technology platforms is the highest.
The second is AI as a feature. Software and service companies that embed machine learning to make an existing product better: fraud detection in payments, demand forecasting in retail, triage in insurance claims. Less glamorous, often more profitable.
The third is AI as an operating lever. Traditional businesses that use AI to change their cost base or their decision quality. Here the investor is not buying technology at all. They are buying the ability of management to absorb it.
Most investment memos I read this year do not distinguish between these three. They should. Each has a different risk profile, a different exit route and a very different relationship to the large cloud platforms that are rapidly becoming the default suppliers of AI infrastructure.
The pandemic as an accelerator
COVID-19 has done something no strategy consultant managed to do in a decade: it has forced boards to digitise in months. Remote work, e-commerce, contactless operations and digital customer service went from “next year’s initiative” to survival. This matters for AI because AI feeds on digital processes and digital data. A company that moved its customer interactions online in April now has the raw material for machine learning that it simply did not have in January.
For investors this means that the addressable market for AI did not shrink during the crisis. It expanded, because the number of companies with usable data expanded.
What I would be careful about
As someone who has run companies and sat on the management board of a private equity fund, I am wary of any technology that is described as relevant to everything. That description is usually true and rarely useful. It tells you nothing about who will capture the value.
The pattern from previous technology waves is instructive. The internet was relevant to everything too. Yet the returns were concentrated in a small number of platforms and in those incumbents who reorganised their business around the new channel. Many of the companies that simply “added the internet” to their name in 1999 did not survive.
I suspect AI will follow a similar curve. The technology will diffuse broadly. The profits will not.
The question I am asking
So the real investment question of 2020 is not whether AI will matter. That debate is over. The question is whether AI should be treated as a sector at all.
My working view is that it should not. AI is becoming a factor that will reprice companies in every sector, in the same way that digital capability repriced retail and media over the last fifteen years. A logistics company with superior forecasting will be worth more than one without it. A bank whose credit models learn faster will win the better customers.
If that is right, then every investor is now an AI investor, whether they have an AI fund or not.
Investment question: Is AI a standalone investment sector or a technology that will fundamentally change the valuation of companies across every sector?
