Key Takeaways:
The shift from chatbots to autonomous AI agents is turning electricity into the most constrained input in the technology industry.
Wall Street is waking up to a brutal physical reality: advanced AI clusters require unprecedented amounts of electricity, and the companies building them are running out of power faster than they are running out of compute. Reliable, cost-effective baseload electricity at scale is hard to come by, and hyperscalers are now racing to lock in nuclear and grid contracts to keep data centers humming 24/7.
"In this agentic AI era, there's a need for inference-optimized, energy-efficient solutions," Durga Malladi, Qualcomm's executive vice president for technology planning and data center, said in an interview. "Our technology is purposely built for distributed, power-efficient, end-to-end systems that can scale across environments."
The Defiance AI and Power Infrastructure ETF (AIPO), which targets the physical nexus of data center buildout including hardware, liquid cooling and localized power generation, has surged more than 40% year-to-date, nearing $800 million in assets under management within its first full year of trading. The ALPS Electrification Infrastructure ETF (ELFY), an equal-weighted play on electrical equipment manufacturers and grid modernization, has gained 30% year-to-date and nearly 50% over the trailing 12 months.
The bottleneck is reshaping where investors deploy capital. Global renewable energy ETFs hauled in more than $3 billion in fresh net cash in April alone, the strongest monthly inflow since January 2021, pushing total clean energy ETF assets to $43 billion. The Range Nuclear Renaissance Index ETF (NUKZ), which tracks the full nuclear value chain with an emphasis on construction and equipment companies, has attracted $835 million in assets as hyperscalers sign power purchase agreements with small modular reactor developers.
The Nuclear Renaissance Gets a Second Act
Uranium-focused funds are also drawing institutional interest. The Global X Uranium ETF (URA), the sector's largest with more than $7 billion in total assets, has attracted more than $850 million in recent net inflows, balancing mining giants with nuclear component integrators. The Sprott Uranium Miners ETF (URNM), a more aggressive commodity play requiring constituents to derive at least 50% of revenue directly from uranium mining or physical holdings, now manages $2 billion and captures maximum price sensitivity to the structural deficit in global uranium supply.
NUKZ takes a different approach, heavily favoring Industrials at roughly 48% of its portfolio and Utilities at 31%, rather than the mining names that dominate URA and URNM. As the world builds new reactors, the equipment supply chain stands to benefit but remains underowned by other strategies, the fund's construction suggests.
Why the Bottleneck Is Structural, Not Cyclical
Coinbase Chief Executive Officer Brian Armstrong argues the scarcity is not temporary. Demand for AI-generated intelligence has no practical ceiling, he said, and as model costs drop and cheap alternatives proliferate, the bottleneck simply shifts upstream to the power and silicon required to run any model at scale. Armstrong expects the market to split within 12 to 18 months, with roughly 80% of workloads migrating to models priced up to 99% below current top-tier options, while the remaining 20% — covering scientific research and high-level orchestrator agents — continues running on frontier systems.
The cost pressure is already visible. Uber exhausted its full 2026 AI budget by April, according to investor Tommy Shaughnessy, illustrating how fast enterprise AI spending can accelerate when metered API pricing replaces flat-rate subscriptions. Open-source models such as DeepSeek V4 now perform within range of top proprietary systems at roughly one-thirtieth the cost, placing a ceiling on what frontier labs can charge.
For investors, the next phase of the AI trade belongs not to the software developers but to the power providers. AIPO trades as a pure-play on data center infrastructure, ELFY offers diversified exposure to grid modernization without mega-cap drift, and the nuclear ETFs provide differentiated access to uranium supply and reactor construction. The ultimate winners in this market regime, the data suggests, will be the companies building and powering the infrastructure behind the code.
This article is for informational purposes only and does not constitute investment advice.
