Huge amounts are being committed to investments in artificial intelligence. Josh Sambrook-Smith, Investment Manager, Global Equities, discusses what this could mean for various parts of the AI ecosystem and where he and the team are seeing opportunities to take advantage.
23 October 2025
It's been a very hot AI summer. So far OpenAI has committed to over a trillion dollars in investment from here to 2030. They've done that across a bunch of different infrastructure providers. We've also had Facebook, Anthropic, xAI, a bunch of other large model makers making not quite as large but very large multi-billion dollar commitments to their infrastructure partners over the next few years.
And so in total, AI investment over the coming years is going to be well in excess of a trillion dollars and it will probably go up. So the immediate winners from this are the obvious hardware makers like Nvidia, AMD but also the wider GPU ecosystem.
So companies who are providing the equipment to make the chips, companies who are providing the software for the chips and companies who are providing the components for the whole box once it's been made.
So there's an emerging debate over profitability. What we've seen recently is that the margins for a lot of this stuff is lower than the market had expected. But this shouldn't be surprising because when you think there are trillions of dollars being invested and there aren't huge amounts of revenue yet, it's very difficult to amortise the cost of all that investment over lots of clients.
So today margins are low and what the industry is betting on is increasing scale and growing revenues to create profitability in the future. So what are revenues? So at the moment, according to the information, AI revenues are over USD 18 billion every year.
And so that includes revenue from the large companies like Microsoft, but also the private companies like OpenAI, Anthropic, xAI, etc. But what's really exciting is the progress. So a year ago, they were doing roughly USD 1 billion of revenue, and we've added nearly USD 20 billion in a year.
So the financial revenue growth is looking pretty good. But also on the product side, the growth has been exponential. So every day, there is a new model being released, there is a new format that is being released to the market.
And if we look at the rate of change, in the next couple of years, I expect the progress of these models to be truly breathtaking. Just as an example, we had OpenAI's developer conference last week. And one of the big takeaways is that they are releasing a social media video generation app called Soar AI. So pretty soon you can look on the App Store and you'll be able to use this video generation application for free yourself.
So how are we invested? Well, we've avoided many of the components companies. A lot of these businesses sell a single component. They are on extremely high valuations, and they're risky. They've only got one customer, it's usually Nvidia, and Nvidia is looking to manage their cost base or change the technologies they use in their devices.
And so whilst these companies are exciting, they've got a lot of growth, they can be dropped very, very quickly, and returns can be hideously bad. We are invested in model makers like Google, for example, with their Gemini app.
And we're also invested in some of the hardware guys like Nvidia, obviously, but also companies like Broadcom who make the custom made high speed chips that are being used by the likes of Amazon, Google, Microsoft, and so on.
We're also heavily invested in the semi-cap space. So the semi-cap companies are the ones who make the equipment that is used to make the chips. And we think the area is interesting because number one, it's got a driver from AI, but also these companies have fantastic business models, and also their long-run performance is independent of AI. So these companies will do well regardless of what happens in the AI space.