Investors seem to have moved beyond debating whether AI matters. Attention is increasingly shifting to a more difficult question: Which parts of the AI ecosystem will ultimately capture the value being created?

In such a fast-moving and dynamic environment, long-term technological progress and shorter-term public market expectations are unlikely to move forward in lockstep. Instead, there will likely continue to be times when investor sentiment overvalues — or undervalues — the evolving business and technological realities. Given that ongoing dynamic, gaining a greater understanding of the AI ecosystem may offer investors more peace of mind amid the volatility and ambiguity.

Beyond the binary

It’s tempting when evaluating the potential impact of a new technology to try to draw up a list of “winners’ and “losers,” based on how likely the new technology may be to benefit or harm existing industries or companies. However, we believe the evolution and proliferation of AI is not dividing companies or industries into binary categories of winners and losers. Instead, many companies are likely to experience both positive and negative impacts from AI.

But while picking winners and losers may be difficult, market narratives around these companies are always in flux. As those narratives and expectations evolve, shifts in investor sentiment often translate into flows of investment into — or out of — a company’s stock.

At some points in time, for example, investors may view the big, well-known technology firms, sometimes known as hyperscalers, as major beneficiaries of AI and be willing to pay high prices for those companies’ shares.

However, at another time, those investors may shift their focus to hyperscalers’ elevated capital spending and opt to reduce their exposure to these stocks. In both instances, investors are considering the same fundamentals but weighing them differently, resulting in volatility and anxiety about the return on capital being spent. Over the past two years, we’ve observed this push-forward-then-fall-back pattern in investor sentiment playing out.

Meet the AI ecosystem

1. Hyperscalers. Large well-known technology platform companies are investing heavily in AI infrastructure. They are major allocators of capital within the ecosystem. An important question for this group is whether returns will ultimately justify their current level of spending.
2. Infrastructure suppliers. These are the companies providing the underlying tools and components required for AI development. They enable the broader build-out of the AI ecosystem.
3. Closed-model providers. These are firms that develop proprietary or so-called frontier large language models.
4. Open-model providers. Organizations developing open-source alternative models from China as well as the US exist as a separate competitive force within the AI ecosystem.
5. Productivity beneficiaries. These are not necessarily technology companies. Instead, they are organizations that use AI to do what they do better, whether that be to increase productivity, lower costs, or improve efficiency.
6. Potentially disrupted companies. These are firms such as software companies that may be negatively affected by AI. The degree to which they may be disrupted remains open to debate, and their outcomes remain unsettled.
7. New business models. These may arise from advances in AI technology and could come from both existing companies and entirely new ones, much as happened during the growth of the world wide web in the 1990s and 2000s.

A pick and shovel approach

If the proliferation of AI seems like a present-day gold rush, it may be worth remembering Mark Twain’s remark: “When everybody is digging for gold, it's good to be in the pick and shovel business.”

The picks and shovels of the AI infrastructure include power generators, turbines, electrical equipment, cooling systems, construction and excavation equipment and more that are required for the massive expansion of computing power. They have no shortage of demand for their products as massive capital spending on infrastructure moves forward.

And moving forward it is. Recent rapid revenue growth from some of the largest AI companies has surpassed consensus expectations, underscoring what we see as the market's underestimation of AI's scale and profit potential. Indeed, we continue to think the market is underestimating the scale of the businesses being built and the profit growth that will follow. Our expectations are moving higher at a pace we would not have believed possible 12 months ago, and we said something similar 12 months ago.

Against this backdrop, we believe that the setup for AI infrastructure stocks remains strong. The forecast for continued growth beyond 2028 seems to be getting better around long-term planning/demand and many key indicators point to continued growth in AI capital spending.

To be sure, some investors may think of industrial AI stocks as a single trade, but that’s incorrect. There have been relative winners and losers over the last 12 months, and dispersion is likely to increase going forward. There are also many cycles within the broader industrial AI cycle: power generators and turbines, electrical equipment, cooling systems, and construction and excavation companies, just to name some of the largest. It’s very possible for one (or more) of these “subcycles” to implode while the broader AI cycle keeps chugging along. The productivity benefits of AI may extend much more widely across business and the economy than they do today, but many investors are not yet willing or able to fully value those potential future benefits today.

For those reasons, we believe that investing in the rapidly evolving AI landscape puts a premium not only on security selection, research, and active management, but also on a holistic understanding of the AI ecosystem. Simply diversifying broadly across a basket of stocks that have exposure to AI may be unlikely to capture the potential of this unique and transformative technology. Instead, active managers with robust research capabilities and a recognition that leadership is likely to shift as technology evolves and market conditions change may be best able to deliver returns while managing risk in this new AI era.