AI Investment Opportunities: Why Investability Matters More Than Scarcity

• Look beyond the obvious AI winners. Rising demand for energy, copper, and critical minerals does not guarantee higher returns for the companies supplying them.
• Find where AI demand translates into investable cash flow. Prioritize companies and markets with attractive valuations, accessible securities, reliable data, and liquidity.
• Look for overlooked opportunities in emerging markets. Improving disclosure, benchmarks, and market access could unlock value in countries supplying critical AI infrastructure.

Artificial intelligence (AI) is one of many examples of an old fundamental problem with investing in technology. Digital technologies can scale quickly. The physical systems that support them cannot. A data center will always need power at a specified location along with things like a grid connection, transformers, cooling, and other prerequisites. It’s a common strategy to invest where the bottleneck is, however, sometimes that thesis fails. For example, you can be the supplier of a critical and growing resource in demand, and still find the economic benefits and public market return of your efforts end up in someone else's hands.

For a global investor, the question is therefore not which resource sits at the bottleneck of AI exposure, instead, which company can turn that constraint into cash flow that allocators can own through an investable asset.

The beneficiaries of AI demand we see in public markets today are not necessarily the companies most responsible for enabling AI. Often, they are simply the companies through which investors can most easily express their investment thesis. If there is no liquid, investable security tied to that cash flow, however, investors cannot easily participate.

By contrast, listed companies with even adjacent exposure can attract capital precisely because they provide a tradable way to own the thesis. Their strong performance may therefore reflect not strictly their economic importance, but also the fact that they are among the few accessible vehicles through which investors can gain exposure. This is the fundamental factor of valuation that investors are mispricing.

A question that isn't as intuitive as most people would think is: When we say “AI demand,” what actually is in demand? Most market participants would probably respond to that with the vague answer of "energy," and, in a sense, this is true. The International Energy Agency's (IEA) updated central case more than doubles global data center electricity demand between 2025 and 2035. In parallel, the IEA expects copper demand for electricity networks to jump 42% to 6.4 million tons by 2035. This confirms what most already know: AI-adjacent sectors like energy and other physical systems matter.

Figure 1 Data center electricity demand and electricity-network copper demand, indexed to 2025 = 100. Data center actuals are rebased to 2025. Gold is the IEA base case; grey dashed lines show the sensitivity range; steel-blue diamonds are IEA electricity-network copper model-year points. Indexing compares growth rates, not physical units or AI-only demand.

It is no secret that data centers critically need energy, and that AI is intensifying a wider grid and electrification race. This race for data centers, energy, and other related infrastructure creates many different markets, but this new resource scarcity is not a single trade. Electricity networks use copper; grid-scale storage uses lithium, graphite, and nickel.

The investment implication is less obvious than the headlines make it seem. A demand increase can be absorbed by new supply, substitution, regulation, or weak project economics. A mineral can matter to a system without making every miner, refiner, or even country an investable winner. A system constraint alone does not necessarily guarantee that there will be a capital markets event that will reward the suppliers of the goods in demand.

This is what we see globally today: those solving the problem of AI demand, whether that be a company refining materials critical for data centers or countries supplying the minerals, are not automatically capturing the economic benefits their strategic importance might suggest. Recent data on electricity-system material demand (Figure 2) make it clear: as the electricity system expands, demand for several critical materials rises with it.

Figure 2 Selected electricity-system material demand, 2025 = 100

However, resource endowment or even control of refining capacity, despite its importance in the supply chain, does not necessarily translate into control of the resulting investment returns. If it did, markets in countries such as China and the Democratic Republic of the Congo would already be capturing far more of the upside from AI-driven demand. The IEA's current copper outlook shows China maintaining the largest projected refined-output base through 2040. Its share of the complete country set rises from 46.8% in 2025 to 51.1% in 2040.

A prudent investor should understand this, while recognizing that it may not remain the status quo. The immediate reality is that global investors can express a thesis such as the AI buildout through developed markets almost instantly. The securities are liquid, the data is standardized, and the benchmarks are familiar. The same cannot always be said for markets closer to the physical bottlenecks.

Nigeria provides a simple example of why this misalignment matters. Venoble’s Nigerian equities: price return versus gross total return index (Figure 3) compares the NGX All-Share Index’s published price path with a gross total-return version that reinvests verified cash dividends. From 2 January 2020 to 5 March 2026, ₦100 grew to roughly ₦733 on the price index, but to ₦978 with dividends reinvested. Put differently, ignoring dividends misses ₦245 of value for every ₦100 invested. In 2025, the gross total-return index returned about 57% in naira; using year-end USD/NGN exchange rates, that translated to roughly 68% in U.S.-dollar terms, more than 50 percentage points ahead of the S&P 500’s 17.9% total return.

Figure 3 Nigerian equities: price return versus gross total return. Common 2 January 2020 to 5 March 2026 window, rebased to 100. End values: price return 733; gross total return 978. Gross total return reinvests verified cash dividends before withholding tax and transaction costs.

Yet that result is far less visible to a global allocator than the equivalent performance in a developed market. This raises a larger question about how much investable value remains obscured by weak data, benchmarks, and market infrastructure. This is only one measurement problem in one market, but it points to a larger one. If better data can materially change the return investors observe, then better benchmarks, custody, disclosure, settlement, research, and market access can certainly change which opportunities they are willing and able to own.

Therein lies the unspoken battle of the AI trade. The competition will not remain only between companies. It will increasingly be between geographies and exchanges competing to turn physical importance into investable value.

Today, the easiest way to express many of these themes is still through developed-market securities, even though important parts of the underlying supply chain sit elsewhere. The opportunity is to identify the markets closest to critical supply-chain bottlenecks as they become more accessible to global investors and begin to be priced accordingly. Investors have already been exploiting this disconnect, not as a future thesis, but as a source of returns today.

AI will change what is scarce. It will not decide who captures that scarcity premium. The true asymmetric opportunity of this trade is not about finding another company with AI exposure, but instead, recognizing where strategic importance is finally becoming measurable, accessible, and investable.