Summarising my current AI thesis – Part 2

How to respond

Across my articles and media commentaries, I have emphasised that preparing for an artificial intelligence (AI) re-pricing is not about timing a top or shorting tech, but rather about protecting capital through discipline, factor balance, and liquidity.

There are five primary strategies to insulate an investment portfolio from an AI re-rating:

1. Diversify away from equity index heavyweights

Broad, passive equity indices (such as the S&P 500 or Nasdaq) are historically concentrated. Passive flows into index funds mean a significant portion of an investor’s exposure is tied to six or seven AI-driven mega-caps.

Rebalance away from broad index funds that leave up to 60 per cent of equity risk tied to the AI trade. Investors might also consider locking in recent gains and diversifying equity exposure across unsung, non-AI themes.

1. Rotate into ‘Unfashionable’ high-quality & value factors

There’s a massive valuation disparity. While AI momentum stocks trade at hyper-extended multiples despite falling cash flows, traditional ‘Quality’ factor stocks have underperformed by levels last seen in 1999, right before the Dot-Com bust.

One idea is to pivot toward true high-quality, cash-generative businesses. Among large and small caps look for:

• High Return on Equity (ROE) and Return on Capital Employed (ROCE).
• High Free Cash Flow (FCF) yields (e.g., >4 per cent) that do not require continuous capital injections.
• Strong balance sheets with low or manageable debt loads.

When a market-wide flight to quality occurs, cash-rich, high-quality, high-Return On Invested Capital (ROIC) businesses could provide structural downside protection.

1. Build up cash reserves (Buffett’s playbook)

Cash is a terrible investment over the long term, but when cash yields are relatively attractive, they provide an option over lower stock prices that pays you while you wait. Warren Buffett’s US$400 billion cash reserve at Berkshire Hathaway suggests having some cash while waiting for sanity to return is completely valid.

Right now, some cash, alongside a reduced exposure to high-momentum AI stocks, could act as both a volatility buffer during a market sell-off and ‘dry powder’ to acquire world-class businesses at steep discounts if a bubble deflates.

1. Private Credit and fixed-income alternatives

To generate defensive yield while equity risk premiums are compressed, you could look to alternative asset classes that are detached or uncorrelated from stock market momentum.

Consider allocating capital into very high-quality, investment-grade private credit and perhaps some high-quality fixed income, noting that in more recent times of trouble, fixed income has moved in concert with falling equity prices. High-grade private credit can provide attractive annual yields with contractual cash returns, offering income shelter while public equity markets might undergo structural re-ratings.

1. Avoid ‘supply-driven’ squeeze points and commodity speculation

While some investors buy peripheral AI supply-chain stocks (like semiconductor equipment makers, power utilities, or memory suppliers), short-seller analysis shows how cyclical supply gluts threaten those trades as well.

Be careful chasing cyclical AI playbooks, for example, memory chipmakers, cooling, or power infrastructure companies, where valuations already price in multi-year hyper-growth. If hyperscalers cut capex budgets, supplier margins and valuations could compress rapidly across the entire value chain.

I currently believe the right approach has to be rooted in capital preservation. Rather than taking a speculative short position, it might make more sense to harvest AI momentum gains, hold a little cash, build income yield through alternative offerings such as private credit, and rotate into more boring, high-quality, cash-generative equities the broader market has temporarily ignored or rejected.