What is changing
Spending on the infrastructure that trains and runs large models has become one of the largest capital cycles in the economy: accelerators, high-bandwidth memory, networking, data-centre power and cooling, and the software that schedules and serves the models. The demand is coming from every large company at once rather than from one industry.
How we express it
Through the layers of that supply chain with pricing power and visible order books, held in liquid, regulated vehicles, rather than through a single company or a private bet on which model wins.
How it is sized
As a sleeve on top of the diversified core, only where a client’s investment policy calls for a growth allocation, with a position limit set in that policy and rebalanced on rules. Many clients do not hold it; for those who do, it is a minority of the portfolio.
What would make it wrong
A sustained fall in the return on model spending at the companies buying the hardware; a step change in efficiency that cuts the compute needed per unit of work; or supply catching up with demand so that pricing power moves to the buyers.
Who it is for
Clients with a long horizon, no need to draw on this part of the portfolio for a decade, and a policy that accepts years of lagging the market in exchange for owning the theme early. It is not for money needed inside five years.
This material is for informational purposes only and is not a recommendation to buy or sell any security. All investing involves risk, including loss of principal.