Surging Small- and Mid-Cap Firms Redraw the Investment Map as AI Drives New Power Demands

Artificial intelligence is broadening its influence well beyond semiconductors and software, accelerating demand for electricity, reshaping infrastructure spending and shifting investor attention toward companies once considered peripheral. Market specialists speaking on this week’s episode of CNBC’s “ETF Edge” said that the physical requirements of large language models and other AI workloads—particularly uninterrupted power and efficient cooling—are enabling a new wave of small- and mid-cap winners, while also testing traditional portfolio strategies.

The sharp price reversal experienced by Bloom Energy illustrates how quickly fortunes can change. After struggling for years to trade above its 2018 initial public offering price, the on-site fuel-cell producer has risen more than 500% since last year as data-center operators rushed to secure dependable electricity. The rally pushed Bloom’s market capitalization past $30 billion and moved the company into the sights of large-cap investors almost overnight.

“Firms that used to sit outside the market’s focus are very quickly moving up the cap table,” said Jennifer Grancio, global head of distribution at TCW Group. She noted that many of the emerging players compete in narrow segments with limited rivalry, allowing revenue and margin trends to improve faster than broad investor awareness.

Energy reliability becomes a gating factor

Electricity stability has emerged as the central constraint for AI expansion. The cost of wind and solar power has fallen sharply over the past decade, but their output remains intermittent. Data centers cannot tolerate that variability, forcing operators to secure constant supplies to avoid costly downtime. As a result, Grancio said, there has been “a huge shift toward nuclear,” including renewed investment in extending the life of existing reactors and developing small modular designs that can be built closer to load centers.

The trend is feeding a growing ecosystem of specialized suppliers, many of which sit upstream from utilities and cloud hyperscalers. Exchange-traded funds focused on nuclear generation and uranium—such as the First Trust Bloomberg Nuclear Power ETF (RCTR), VanEck Uranium and Nuclear ETF (NLR), Themes Uranium & Nuclear ETF (URAN), Range Nuclear Renaissance Index ETF (NUKZ) and Global X Uranium ETF (URA)—have become popular vehicles for gaining exposure to that theme.

Utility regulators and policy makers are also reassessing baseload capacity. According to the U.S. Energy Information Administration, nuclear plants supplied roughly 19% of domestic electricity in 2023, a share that could rise if life extensions and modular reactors come online as planned.

Inside the data center: cooling and power management

Even with reliable generation, capacity inside the data center remains tight. The rapid growth of graphics processing units and high-performance chips has made cooling and power management the next set of chokepoints. Grancio said investors are gravitating toward companies that rank first or second in their niche, especially when only a handful of viable alternatives exist. The resulting concentration can produce substantial operating leverage, but it also magnifies the impact of execution missteps.

Imagem: Internet

Such market structures are fueling interest in actively managed ETFs. Passive indices eventually incorporate successful newcomers, yet active managers aim to identify them earlier and hold positions through multiple growth phases. TCW and other firms argue that a research-driven approach is better suited to navigate sectors where technological change and capital-spending cycles move quickly.

Volatility and portfolio construction

Jan van Eck, chief executive of VanEck, cautioned that parts of the current AI supply chain include “small, financially weak companies” whose prospects are tightly linked to electricity demand. That dependence can translate into sharp price swings. Van Eck noted that his firm’s nuclear ETF traded at what he called “nosebleed levels” in 2023 before retreating to valuations he considers more reasonable for new buyers.

The message from ETF sponsors is that no single AI-driven theme should dominate an investor’s allocation. Broad diversification, active rebalancing and clearly defined risk limits are viewed as essential, particularly as more investors bring AI exposure into portfolio construction in a deliberate way over the next two years. Maintaining discipline, the panelists said, can help participants avoid chasing momentum at peaks or capitulating during inevitable pullbacks.

For now, the strongest takeaway is that artificial intelligence is transforming industries far beyond its software roots. Companies supplying round-the-clock power, advanced cooling and other infrastructure components are climbing market-cap rankings at record speed, providing fresh opportunities—and fresh risks—for investors willing to look past the traditional technology leaders.

Crédito da imagem: CNBC