Beyond the MANGOS

• The AI theme is moving beyond asset-light technology platforms and into a more capital-intensive infrastructure cycle.

• Unlike earlier internet-era winners, AI growth requires substantial ongoing investment in compute, power, cooling, networking and data centres.

• This creates broader opportunities across companies supplying the infrastructure needed to support AI adoption.

• Our focus is not on avoiding AI, but on identifying where the economic benefits ultimately flow and whether valuations already reflect the opportunity.

The rise of MANGOS reflects a shift in market thinking, but it also marks an important change in the economics of technology leadership. FAANG represented the internet and smartphone era, where many dominant businesses benefited from highly scalable, asset-light models. Once the platform was built, adding another user, search query, advertising impression or digital subscription often carried a very low marginal cost.

MANGOS is different. The AI ecosystem is far more capital intensive. Training and running large models requires data centres, advanced semiconductors, networking equipment, cooling systems and reliable electricity. Compute is not free: each additional query, model update or enterprise deployment consumes processing capacity and power. Recent estimates suggest global data centres may require trillions of dollars in capital investment by 2030 to keep pace with demand, reinforcing that AI is not simply a software story but an infrastructure cycle as well.

For investors, this creates opportunities beyond technology alone. The businesses supplying critical infrastructure may prove just as important as those creating the applications.

Rather than attempting to predict which AI model developer or platform company will ultimately dominate, our portfolios focus on businesses that can benefit from the broader adoption of artificial intelligence and the investment required to support it.

For example, we hold Vertiv, a company that provides critical power and cooling infrastructure used in modern data centres. As AI workloads become larger and more energy intensive, demand for the physical infrastructure supporting those systems continues to grow.

We also own businesses such as Emerson Electric and Xylem, which benefit from rising investment in industrial, electrical and water infrastructure. Whilst these companies may not generate the same headlines as the MANGOS, they provide many of the essential systems required to support the expansion of digital infrastructure across the global economy.

This reflects an important aspect of our investment philosophy. We do not need to own every company at the centre of a popular narrative. Instead, we seek opportunities where expectations are more modest and valuations are more reasonable, while still maintaining exposure to long-term themes driving economic growth.

The MANGOS may prove to be extraordinary businesses over time. However, successful investing is not simply about identifying great companies. It is about identifying situations where future outcomes are better than current market expectations. History offers an interesting lesson. Some of the most transformative innovations have been disappointing investments at the peak of investor enthusiasm.

Railways transformed transportation. The internet changed how the world communicates. Fibre-optic networks rewired the global economy. In each case, the technology ultimately succeeded beyond most expectations. The investment returns were often less spectacular.

The reason was simple: competition and capital.

While the leading US models remain at the frontier of capability, lower-cost alternatives continue to emerge and the gap between frontier and open models is narrowing (see chart below). The ability of open models to remain competitive despite substantially lower pricing may make it difficult for AI providers to earn the premium returns required to justify the capex investment.

None of this means AI will fail. Quite the opposite.

Artificial intelligence will almost certainly create significant economic value. The more interesting question for investors is whether that value ultimately accrues to the companies developing the models, the businesses supplying the infrastructure, the enterprises adopting the technology, or simply their customers.

To extend the fruit analogy, the mangoes may well be excellent, but will consumers prefer a mandarin at one-tenth the cost?

The AI revolution may create significant value over the coming decade. Our focus is not on avoiding the theme, but on identifying where the economic benefits ultimately flow. Sometimes that will be the company selling the mangoes. More often, it may be the businesses supplying the orchard, irrigation system and delivery trucks.

In our view, successful investing in the AI era is likely to require looking beyond the MANGOS.