AI’s Long-Term Potential: More Upside Than Downside

As a follow-up to the AI-Enabled Productivity theme introduced in our February release of the 2025 Capital Market Assumptions (CMA), we’re taking a closer look at how this thesis is evolving.

Our original view suggested that the rapid advancement and adoption of AI could help offset productivity headwinds stemming from demographic shifts. While short-term implementation challenges persist, we remain confident in AI’s long-term potential to drive meaningful productivity gains. Recent market developments also warrant renewed attention.

Demographic challenges continue, exacerbated by changing immigration policies. At the same time, AI development is accelerating. For example, China’s announcement of DeepSeek R1 initially triggered a brief U.S. stock market dip, but the drawdown was short-lived, with competition viewed as a catalyst for innovation and faster AI progress.

AI is also being deployed in robotics (e.g., robotaxis, humanoid robotics)1 and agriculture, with China rolling out over 251,000 crop protection drones in 2024.2 These advancements could have a meaningful impact on food inflation and overall productivity.

We continue to view AI as having an asymmetrical profile — offering greater potential for upside than downside. However, concerns about job displacement remain. The timing and scale of AI’s productivity impact will vary across developed, emerging, and frontier economies, depending on each region’s technological readiness and workforce composition. Economies that embrace change may benefit from faster adoption and growth.

The Supply Chain Behind AI: Why Critical Minerals Matter

Trade tensions and concentrated mineral supply chains present strategic risks to AI scalability. China’s willingness to restrict critical mineral exports as a non-tariff trade barrier is increasing uncertainty in the global supply chain. For example, following tariffs imposed earlier this year, China’s exports of rare earth magnets dropped 74% in May compared to a year earlier — with exports to the U.S. falling 93%.3 These magnets are essential for graphic processing units (GPUs) used in AI model training and for small motors and actuators in robotics.

While the recently agreed U.S.-China trade framework aims to address these concerns,4 dependence on non-domestic sources means uncertainty will likely persist until supply diversification improves.

This issue was discussed at the latest G7 meeting in Kananaskis, Canada, where an action plan was agreed upon, though a near-term solution remains out of reach.5 China, Indonesia, and the Democratic Republic of the Congo, currently control 86% of the global refined critical minerals supply, up from 82% in 2020.6 About 90% of supply growth came from single-country sources — Indonesia for nickel, and China for cobalt, graphite, and rare earths.7 Exhibit 1 shows how China dominates in mineral refining.

Critical mineral supply also interacts with our other CMA themes: Navigating the Energy Transition relies on critical minerals for batteries and other technologies; and Globalization: Bent not Broken, explores trade, supply chains, and related risks and opportunities.

EXHIBIT 1: CHINA DOMINATES CRITICAL MINERAL REFINING

Percentage share of global refined material supply.

Source: Northern Trust Asset Management, Macrobond, U.S. Bureau of Economic Analysis (BEA). Q = quarter; GDP = Gross Domestic Product. Data as of 6/30/2025. Historical trends are not predictive of future results.

The Infrastructure Opportunity Behind AI’s Ascent

Infrastructure, both listed and private, remains a potential winner under our AI-enabled theme. Infrastructure has long been a favored option of global investors due to the sector’s high barriers to entry,  inflation-linked cash flows, and equity-like returns with lower volatility. This theme, as well as Navigating the Energy Transition, further supports the case for infrastructure investment.

In the U.S., deregulation, the push for technological dominance (e.g., data centers, generative AI), and strong support for onshoring all act as tailwinds for the asset class. Today’s generative AI needs represent a game-changer for the utility and power-generation sectors. With many end-users seeking large-scale campuses, demand for power and data capacity is expected to remain robust.  AI-driven data center power demand is projected to exceed 123 gigawatts by 2035 — a 30x increase from 2024.8 Utilities, midstream energy, and nuclear power are poised to benefit from growth in data centers, manufacturing, and electrification of the economy.

Outside the U.S., many countries are accelerating domestic infrastructure investments. For example, Germany announced the creation of a €500 billion infrastructure fund to support investments across energy, transport, and digital infrastructure. Globally, investments in AI-related data center capacity continue to grow, with one estimate putting it between US$3 trillion and US$8 trillion by 2030.9

AI’s Uneven Impact on Growth and Inflation

AI adoption has the potential to drive meaningful productivity and economic growth in countries able to capitalize on this theme. However, the magnitude of this impact remains difficult to project. Its potential is asymmetrical and skewed to the upside. For economies unable to leverage AI, economic growth may be more limited.

The effects on inflation are less clear. AI could lead to lower long-term inflation if it enables cheaper production and contributes to job displacement. More people competing for fewer jobs could lessen wage pressure, which may drive inflation lower.

AI, Energy, and Globalization: A Converging Investment Story

As AI drives up power demand, especially from data centers and generative AI,  energy supply must keep pace. In the U.S., traditional sources like coal and petroleum are declining,10 while renewables face policy challenges. This growing demand is leading to the first major increase in the power load in a decade and prompting decisions to extend the lifespan of existing power plants.

A recent example is Meta’s 20-year deal with Constellation Energy to purchase energy from an Illinois nuclear power plant.11 Previously slated for decommissioning, the agreement extends the plant’s life by providing data centers more than 1100 megawatts of power.

As noted earlier, concerns about critical mineral availability and China’s use of critical minerals as potential non-tariff leverage in trade negotiations, also tie into Navigating the Energy Transition and Globalization: Bent not Broken CMA themes. These materials are essential for batteries and other energy infrastructure technology.

Special Considerations China’s Strategic Role in AI and Market Leadership

One of our “Special Considerations” in the 2025 CMA was China, specifically its economic size and trade influence. While that consideration focused on domestic consumption and internal stimulus, China remains relevant to the AI-Enabled Productivity theme through its role in trade, trade barriers, and competition for AI leadership. The U.S.’s ability to maintain AI innovation and leadership could also influence our Market Concentration topic in our CMA, particularly regarding continued U.S. stock dominance.