Artificial intelligence (AI) has reshaped financial market leadership, capital deployment, business operations, and how investors assess long-term growth. For long-horizon investors, the question is no longer whether AI matters, but how its evolution may reshape the investment landscape. Here, we explore the current wave of AI excitement among investors and the economy, and how markets are pricing this story.
How AI became a powerful technology
AI may appear to have emerged suddenly, but today’s progress reflects decades of research. AI was first defined as a formal research field in the 1950s, but limited computing capacity slowed development. Today’s generative AI models are built on “deep learning,” a subset of machine learning inspired by the human brain. Deep learning concepts date back to the 1940s, but they required far more computing power than earlier systems could provide.
The scaling up of cloud computing in the 2010s has helped unlock models that can read handwriting, recognize images, and ultimately power today’s systems such as ChatGPT, Claude, and Gemini. Their success, however, depends on enormous capital demands for development, deployment, and operation of data centers.
The enablers of AI
Since generative AI became mainstream, adoption has moved quickly, with more than two billion monthly users combined across major platforms.1 Capabilities are advancing on timelines of weeks rather than years, supported by a surge in AI infrastructure spending. US hyperscalers expect to spend $730billion this year largely on AI projects,2 while Gartner expects 2026 AI-related spending to exceed $2.5 trillion globally. In our analysis, most companies have articulated some form of AI ambition.
Most of this spending comes back to computing demand. According to Epoch AI, compute needs for leading-edge models have roughly doubled every six months over the past decade. After a model is trained, ongoing “inference” costs are incurred each time the system is used, and these demands are rising with broader adoption and the emergence of agentic AI.
To meet this demand, hyperscalers, neoclouds, and data center builders are racing to secure compute capacity across four core verticals: commodities, semiconductors, data center infrastructure, and energy.
Energy may be the most salient constraint and the fourth vertical, in our view. The International Energy Agency expects global data center electricity consumption to grow 15% annually through 2030 — four times faster than all other sectors — and rise to 3.0% of total electricity use.3 In developed markets, data centers are expected to contribute 20% of total electricity demand growth over the same period. As grid capacity tightens, builders are turning to “bring-your-own-generation” models using sources such as natural gas-fired turbines or nuclear energy, both of which can have 5- to 7-year lead times.
From infrastructure build-out to fundamentals
So far, market exposure to AI has been captured in semiconductors and data center-related businesses. This trade has been supported by rapid computing capacity expansion to increase AI capabilities/use cases and rising consumer and business adoption.
We expect this to continue, but with more nuance. Markets have recently begun differentiating between companies based on balance sheet strength, capital discipline, and paths to monetization. Where AI-related names previously moved together in an “AI beta trade,” we believe markets are now placing greater emphasis on fundamentals, capital structure, and the economics of the build-out.
AI infrastructure also faces physical constraints, with power availability, grid capacity, and local opposition emerging as bottlenecks. These frictions may not stop the AI build-out, but they are likely to reshape its economics and the profile of long-term returns.
Productivity and growth
The long-term macroeconomic case for AI rests on productivity. Generative AI can assist, automate, and augment work, especially in information-intensive tasks. Early use cases, including software development, suggest meaningful gains are possible, though estimates vary widely. Over time, AI is likely to replace some tasks and transform many others, while still leaving a substantial role for labor in our view.
If AI adoption follows prior technology cycles, productivity gains may be incremental and uneven at first, then broader and more powerful over time. For developed economies facing aging populations and labor shortages, AI’s potential as a labor-augmenting technology is particularly relevant. Over the long run, higher productivity should support average growth rates and exert disinflationary pressure, in our view.
Disruption and dispersion across sectors
As AI capabilities improve and costs fall, attention is shifting from who builds it to who may be disrupted by it. Software, professional services, financial intermediation, and other information-heavy industries are increasingly exposed as AI tools transform or substitute some business models. Price action in the first quarter of 2026 is an example of some of these fears playing out, where software companies were under selling pressure resulting from recent AI advancements.4
Still, not all sectors are equally vulnerable. Activities requiring trust, accountability, explainability, complex regulation, proprietary data, embedded switching costs, or highly variable real-world interactions may retain more human involvement and face less near-term disruption. This should increase valuation dispersion within and across sectors, making diversified exposure important for long-term investors.
Capital allocation beyond technology
A key shift ahead may be the diffusion of AI opportunity beyond a handful of major technology names. Early leadership has been highly concentrated in the US and in semiconductor names, but other approaches are emerging elsewhere, such as China.
National AI strategies are likely to continue defining the playing field. Risks include overinvestment, regulation, geopolitical fragmentation, social backlash, changes in labor bargaining power and income distribution, and evolving norms around data ownership and intellectual property. These forces may shape regional AI strategies and influence economic outcomes and political responses over time.
What this could mean for long-term investors
We believe AI remains in its early economic stage. For long-term investors, we suggest these strategic implications:
- We are focusing on long-term productivity and growth effects, particularly in economies positioned to benefit from AI adoption.
- We expect greater dispersion in returns as disruption broadens beyond the technology sector.
- We manage concentration risk and remain alert to second-order consequences.
In our view, navigating the AI evolution will require patience, perspective, and a willingness to look beyond today’s most obvious beneficiaries. We remain attentive to how the investment landscape changes and which companies are best positioned to capture value in the new environment.
