Diversifying away from the AI bubble: sectors and regions

What does diversifying away from the AI bubble mean?

It means holding more assets whose prices follow other drivers than AI spending. The goal is a portfolio that keeps its value better if AI prices fall. You can still hold AI names, and you hold them as one part. The EQOS Asia AI study records valuations up ×4 in 18 months (April 2026), which shows how fast a single theme can grow in a portfolio.

Diversification works through weights, not labels. A portfolio with 20 holdings can still depend on one theme if 15 of them respond to AI demand. Count the share of the portfolio that responds to AI, and aim to lower that share. The number tells you more than the count of holdings. Track it each quarter, next to the date and the prices that moved it, and you build a record of how your dependence changes.

How do you find hidden AI exposure before you diversify?

Look inside your funds. An index fund weighted by company size holds large AI names in proportion to their size, so a rise in those names raises their weight. Open the fund page, read the top 10 holdings with their weights, and add them to your direct holdings. The same names can appear 3 times: in direct shares, in index funds, and in thematic funds.

Follow 4 steps: list every account, look through each fund, shift the weights, and recheck on a set date. The look-through step shows the real picture. The shift step needs a target, such as a ceiling on the AI-linked share. The recheck step keeps the work from fading after one afternoon, because the weights drift every time prices move.

Which sectors reduce dependence on the AI bubble?

Sectors that sell to households or follow slower cycles tend to depend less on AI spending. Healthcare and consumer staples are common examples, since their demand rests on needs that exist in any market. Check each choice for a hidden link: a fund can hold technology suppliers, and a sector label alone does not remove the exposure.

Some sectors link to AI through a different door. Utilities and energy firms may supply the power that data centres use, so their prices can rise and fall with AI build-out plans. These sectors still diversify against some risks, and they connect to AI through demand for power. Read the holdings before you rely on the label.

Which regions help diversify a portfolio away from AI stocks?

Regions differ in their sector mix. Markets with more banks, consumer firms, and industrial firms depend less on technology than markets with large chip and software firms. For investors in Singapore, ASEAN, and Hong Kong, a regional fund can add variety. Compare its top holdings with your AI list, since some regional funds hold chip makers and data-centre suppliers.

Currency adds a second layer. Holding assets in other currencies spreads the risk further, and it adds exchange-rate swings to the result. Decide which currency your goals are set in, then size foreign holdings to match. A region that adds variety in business risk can add risk in currency, so weigh both. Many readers keep the home currency for the core and add foreign holdings in steps.

How much diversification away from AI stocks is enough?

Illustration: a portfolio holds 50% in AI-linked assets, counting direct shares, funds, and income. A fall of 30% in those assets costs 15% of the portfolio. Moving the AI-linked share to 30% cuts the same fall to a 9% hit. The change in the share tells you what the shift buys.

Pick the AI-linked share by the loss you can carry, then work backwards. If a 15% portfolio loss changes your plans, the 50% share is too high. If 9% fits, 30% works. Write the limit down, and let rebalancing rules bring the portfolio back to it when prices drift.

What does it cost to diversify away from the AI bubble?

The main cost is the AI upside you give up if the boom continues. A smaller AI weight earns less when AI prices rise. Add trading fees, taxes on gains from the shares you sell, and the fees of any new funds. These costs are real, and each is visible in advance.

Compare the cost with the protection. Selling part of a large gain to cut the AI weight locks that gain in, and the tax on it is the price. Phasing the move over several dates spreads the cost and the timing risk. A plan with 3 steps over a year is easier to follow than a single large trade.

How does EQOS Asia scoring test a plan to diversify away from the AI bubble?

The EQOS Asia AI study scores the enterprise AI market at φ(S) 0.337, a fragile zone where a shock can tip the system, with bifurcation as the verdict. A diversified portfolio should hold up in a soft landing and in a break. Scoring both shows how much each shift of weight is worth.

The Λ simulator gives a first estimate of your case in 10 minutes. For larger portfolios in Singapore, ASEAN, or Hong Kong, the EQOS Asia pricing page lists formats from a 30-minute pre-diagnostic to a full structural diagnostic.

Frequently asked questions

About the author

Stéphane Jambu

President, EQOS Forecast Asia

Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile