Every major technology cycle produces the same investment dilemma: a genuine long-term structural shift is underway, but near-term valuations may have run ahead of fundamentals, and narratives have attracted capital that does not fully understand what it is buying.
The artificial intelligence investment theme in 2026 is no different. The technology is real. The economic implications are real. The question is whether current valuations reflect those fundamentals or whether they price in a future that may take longer to materialise — or may not arrive in the form currently anticipated.
The case for AI as a structural investment theme
The economic case for AI is serious and should not be dismissed. Several factors distinguish this cycle from previous technology bubbles:
Revenue is real. Unlike the dot-com era, where many high-valued companies had little or no revenue, the major AI infrastructure companies — chipmakers, cloud platforms, enterprise software providers — are generating substantial and growing cash flows. AI investment is being driven by corporate capex from profitable businesses, not venture speculation alone.
Productivity gains are measurable. Across software engineering, legal work, customer service, drug discovery and multiple other sectors, measurable productivity gains from AI deployment are being reported. This is not purely speculative — businesses are seeing genuine cost reduction and capability improvements.
The infrastructure investment cycle. The scale of data centre construction, power infrastructure build-out and chip manufacturing investment underway globally is comparable in scale to previous transformative infrastructure cycles — electrification, the internet. Infrastructure investments of this scale create durable economic activity.
Competitive dynamics are accelerating adoption. Once AI provides measurable competitive advantages in specific industries, companies that fail to adopt face a real risk of competitive displacement. This creates structural demand that is not purely discretionary.
The case for caution on valuations
Against this, several concerns merit serious weight:
Concentration risk. The AI theme has been disproportionately concentrated in a small number of US mega-cap stocks. A handful of companies account for a very large share of US equity indices. Investors in passive equity funds have substantial AI exposure without necessarily realising it — and without having made an active decision about how much concentration is appropriate.
The monetisation gap. There is a difference between the infrastructure investment phase — which benefits chipmakers and cloud providers — and the monetisation phase — where AI generates returns in end-user applications. The transition between these phases is not guaranteed to be smooth or rapid.
Energy and resource constraints. AI data centres consume enormous quantities of power. The intersection of AI infrastructure demand with energy transition requirements is creating genuine constraints in some markets. The cost of power is a material input cost for AI services.
Regulatory risk is increasing. EU AI regulation (the AI Act) is being implemented. US regulatory scrutiny of both AI applications and the competitive dynamics of AI platform markets is growing. This does not necessarily reduce long-term value, but it adds uncertainty.
Alternative technology cycles as a reference point. The internet was a genuinely transformative technology. It also produced a bubble in 1999–2000 that took 15 years to recover from — even though the underlying transformation was entirely real. Timing matters in investment, and being right about the long-term trend does not insulate against near-term losses.
Valuation: how extended is the market?
By mid-2026, the price-to-earnings ratios of the largest AI-exposed US technology stocks are at levels that price in years of continued high growth. The market is not pricing in a scenario where AI deployment disappoints, where regulation bites harder than expected, or where the competitive dynamics of AI shift value away from current market leaders toward open-source alternatives or non-US providers.
That does not mean the market is wrong. Technology companies with genuine platform advantages often sustain high valuations for longer than sceptics expect. But it does mean that the margin of safety for investors entering at current valuations is thin — and that the probability of losing money over a 3–5 year horizon is meaningfully higher than it would be for a diversified global equity portfolio at average valuations.
For investors who established AI exposure 2–3 years ago and are now significantly in profit, the question of rebalancing is live. Allowing an initial 3% allocation to drift to 12% after a period of strong performance means the portfolio-level risk is substantially different from what was originally intended.
How to get AI exposure without overconcentrating
For HNW investors who want genuine exposure to the AI theme without making a concentrated bet:
1. Recognise what you already have. Most diversified equity portfolios already have substantial AI exposure through US mega-caps. Before adding more, understand what you hold.
2. Consider the whole supply chain. AI exposure does not mean only investing in AI software companies. Chipmakers (particularly those making specialised AI chips), energy infrastructure companies supplying data centres, and industrial companies benefiting from AI-driven manufacturing efficiency all represent exposure to the theme through different risk/return profiles.
3. Dedicated AI funds vs. thematic ETFs. Several specialist AI and technology funds exist. Assess how concentrated the underlying holdings are — many have high overlap with the few mega-cap stocks you likely already own.
4. Private market access. For eligible HNW investors, some of the most compelling AI investment opportunities remain private. AI infrastructure funds, venture strategies focused on enterprise AI applications, and secondaries funds provide access to the theme outside public markets. These come with illiquidity risk, so position sizing is important.
5. Do not try to pick the winner. The history of technology cycles suggests that the long-term beneficiaries of transformative technology are sometimes not the leading companies of the initial phase. A diversified approach that does not depend on one company sustaining its current dominance is more robust.
The international dimension: AI and portfolio currency exposure
For internationally mobile investors, AI investment carries an implicit currency dimension. The largest AI-exposed companies are overwhelmingly US-listed and US-dollar-denominated. For investors whose costs and spending are in euros, sterling, UAE dirhams or other currencies, a concentrated AI allocation magnifies USD exposure.
Whether USD concentration is desirable depends on your circumstances — but it is worth making the decision explicitly. Investors who want AI theme exposure without doubling their USD concentration can look at European industrial companies benefiting from AI-driven efficiency, or at global technology funds with more diversified geographical exposure.
Our view
AI is a genuine long-term investment theme. The structural case is well-founded. Current valuations in parts of the market warrant caution — not avoidance. A measured allocation, thoughtfully structured across the supply chain, sized appropriately relative to existing holdings, and reviewed regularly, is a sensible approach for most long-term investors.
What is not sensible is either ignoring the theme entirely or concentrating a disproportionate share of wealth in the handful of stocks that happen to be most visible right now.
Frequently asked questions
Should I sell my AI holdings given the valuation concerns?
Not necessarily. The question is not whether to hold any AI exposure — a diversified portfolio will naturally include some — but whether your current concentration is appropriate for your risk profile and time horizon. If AI-related stocks now represent more than 10–15% of your portfolio, reviewing that position size is prudent. Rebalancing gradually over time (rather than all at once) reduces timing risk.
Is there an AI ETF that provides broad exposure without excessive concentration?
Several thematic AI ETFs exist, but many have very similar underlying holdings to a US large-cap index — the same handful of mega-cap stocks. Look carefully at the top 10 holdings and their combined weight. A genuinely diversified AI ETF would include hardware, power infrastructure, data centres, robotics and industrial AI — not just consumer-facing software companies.
What is the risk if AI regulation becomes more restrictive?
EU AI Act implementation is already imposing compliance costs on AI system providers. More severe restrictions — for example, on training data use, on AI in specific sectors, or on data handling — could impose material costs and delays on AI deployment. The risk is higher for consumer-facing AI applications than for enterprise productivity tools, which tend to attract less regulatory attention.
How does AI investment fit into an offshore portfolio bond?
For internationally mobile investors holding investments within an offshore bond wrapper, AI-theme exposure can be gained through the funds and ETFs available on the bond platform. The tax treatment of gains within the bond — where growth rolls up without annual taxation — can be particularly beneficial for volatile assets that may experience large gains and losses. Discuss fund selection with your adviser in the context of your overall allocation.
This article is for general information and educational purposes only. It does not constitute a recommendation to buy or sell any specific investment. The value of investments can fall as well as rise and you may get back less than you invest. Past performance is not a reliable indicator of future results. Contact Global Investments to discuss how international equity and alternative investment exposure is structured within your portfolio.
This article is for general information only and does not constitute financial, legal or tax advice. Rules, prices and regulations change; verify current requirements with a qualified adviser before acting.