Investments  

Why infrastructure investing in AI is 'like selling shovels to gold miners'

Why infrastructure investing in AI is 'like selling shovels to gold miners'
AI is the fastest-growing sector for investors. (This Is Engineering/Pexels)

British start-ups in the artificial intelligence and tech space are burgeoning, giving fund managers and investors plenty of opportunities to back British enterprises, venture capital trust managers have said.

Simon King, fund manager of the Octopus Future Generations VCT, said despite wider concerns over the UK economy, grassroots businesses were burgeoning, particularly in the technology space. 

He said: “The UK remains the start-up capital of Europe with nearly twice as many unicorns as Germany (the next in line) according to the 2022 State of European Tech report.

Article continues after advert

"We also have an amazing set of educational institutions that attract talent and companies from around the world; Amazon, Apple and Microsoft all have research centres in Cambridge and Meta and Google have them in London.

"Our universities are also creating hundreds of spin-outs each year across the UK."

He said that, as a result of this growth, the team was "seeing great opportunities from across the UK" and from European companies that have set up company offices in the UK to take advantage of the skilled people they can find here.

But while AI is the "hottest sector" in tech, King said there were areas that looked overcooked, others where developments are moving almost too quickly and other areas yet where there are strong investment opportunities. 

He explained that some companies, such as chip provider Nvidia, were seeing "sky-high" valuations. Others were moving "at such a pace from one moth to the next" that things are changing almost too rapidly to keep on top of developments.

But King added: "We think about the segment in layers: applications (uses of AI), models (the AI algorithms themselves), software infrastructure (tools to help build and deploy AI models) and hardware infrastructure (the silicon that runs AI models).

"There are pros and cons to each, but we currently feel that applications is the hardest area to invest in because it is evolving so quickly."

However, the team had already somewhat invested in this area as many of the portfolio companies are making use of AI models to upgrade their products, for example an investment in language learning app portfolio company Memrise, that has a new chatbot to help you learn languages, Membot.

But King said, "Overall, models require vast amounts of money to train, and we remain sceptical about their defensibility unless you have something special.

"For example, we have one company in this space, UnlikelyAI, that has a very unique approach. Infrastructure investing in this space is like selling shovels to gold miners, and a great many people will be searching for AI gold over the next few years.

"This is an area we are focusing a lot of time on and have made several investments in already, such as TitanML, a business that helps companies to fine tune their own models for specific applications.”