When everything is the AI trade, what does diversification mean?
Bloomberg’s Big Take this weekend (Lu Wang, 20 September 2026) describes a problem that the world’s largest allocators are only now confronting: the diversification they rely on has quietly become an illusion. AI infrastructure companies now make up roughly 40% of the S&P 500 by Goldman Sachs’ estimate; AI accounts for nearly half of investment-grade bond issuance this year and 87% of venture funding, per Apollo. The CIO of New York City’s $327bn pension system told Bloomberg he now turns down funds simply because he can no longer tell when his AI exposure is “too much”.
That last figure, 87% of venture dollars, is the one that concerns us most, because it describes our own asset class. When nearly nine in ten venture dollars chase one theme, three things follow. Valuations in that theme detach from fundamentals. Capable founders building outside it struggle to get funded, however good the business. And any venture portfolio assembled by following the flow of capital ends up as a single bet dressed up as a diversified one.
There is an older principle at work here, and it predates AI by a long way: a portfolio of small companies is only diversified if the companies are exposed to different things. Sector labels are not enough. A medical device business, a materials company and a security hardware firm can each be built with a modest amount of capital, each rise or fall on its own customers, regulators and supply chains, and each be largely indifferent to what happens to the price of compute. A dozen software companies that all sit on the same three model providers are, for practical purposes, one position.
The point is not that AI is over-hyped. Plenty of good companies would not exist without it. The point is the one Bloomberg’s sources make: exposure to a theme is a risk factor in its own right, and it has to be measured deliberately rather than discovered after the fact. Morgan Stanley’s Lisa Shalett puts it plainly: railroads, the internet, “at some point you’re going to have enough infrastructure… it’s just a question of when”.
For an early-stage investor, the practical test is simple. If the model providers cut their prices by 90% tomorrow, or raised them tenfold, which companies in a portfolio would be better off, which worse, and which wouldn’t notice? A portfolio where the honest answer to the last question is “most of them” is diversified. One where it is “none of them” has made a single call, whatever the sector mix says.
Read the Bloomberg piece: https://www.bloomberg.com/news/features/2026-09-20/ai-boom-is-making-diversifying-investments-tough-for-wall-street
Albion Trinity Capital invests in early-stage companies across healthcare, technology and business services. This article is for information only and is not an offer or solicitation of any investment.