Brunner managers question whether AI’s vast build-out can earn its keep

The Brunner investment trust’s managers argue that the AI build-out’s projected price tag is racing ahead of the revenues needed to justify it. Their warning is not that AI lacks value, but that a transformative technology can still make for a poor investment.

Watch Desk analysis

What happened

In an interview with Trustnet, co-managers Julian Bishop and James Ashworth challenge assumptions about how long today’s AI spending can continue. The article says hyperscalers are forecast to spend about $1.7 trillion on the build-out next year, with cumulative spending approaching $7 trillion by 2030.

Bishop argues that spending on that scale would need much greater revenue to make financial sense. For the projected $7 trillion outlay, he estimates the industry would need around $5 trillion in revenue. Those are the managers’ calculations and forecasts, not settled figures for future spending or sales.

The managers point to chip depreciation and the cost of building and running data centres as part of the bill. They also argue that the current wave of investment may be closer to heavy industry than the lighter-cost technology businesses investors often have in mind. Brunner still has AI exposure through holdings including TSMC and Alphabet, but the article says its allocation is below that of a typical market index. Read the Trustnet interview.

Why it matters

AI investment is often discussed through chip demand, model launches and data-centre plans. The Brunner managers’ case puts the less glamorous question on the table: how much revenue must eventually flow back to support all that capital spending, especially when equipment wears out and facilities cost money to operate?

Their argument is an investment view, not a verdict on whether AI is useful. As Ashworth notes in the interview, a technology can create enormous value for society without delivering equally attractive returns to the companies and investors financing it. That distinction matters to anyone treating the AI boom as a guaranteed winning trade.

Our read

This is a useful challenge to the assumption that demand alone justifies the build-out. But the managers’ estimates depend on forecasts about future spending and revenue, neither of which is a receipt already in the till. Read this as a test of the investment case, not a forecast we can stamp “settled”.

What to watch

• Whether major AI builders keep raising spending plans, or start to rein them in.
• Whether AI-related revenues grow enough to support the infrastructure bill.
• How investors account for chip replacement and data-centre operating costs.

Discussion spark: If AI becomes hugely useful but the returns fail to match the infrastructure bill, was the investment boom still a success?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.