Against the backdrop of an intensifying arms race in AI computing power, a growing consensus is taking shape in capital markets: the ultimate ceiling of AI is increasingly being defined by electricity. As technology giants such as Google (GOOG.US / GOOGL.US), Meta (META.US), and Amazon (AMZN.US) continue to ramp up investment in AI infrastructure, what is truly being amplified behind the scenes is a long-term competition for stable, low-cost power resources.

From “Computing Power First” to “Power Comes First”

Over the past year, the AI rally has not only pushed core computing names like NVIDIA to repeated highs, but has also ignited strong performance across power and energy-related assets. In the U.S. power sector, both traditional utilities and companies involved in nuclear energy, natural gas, and renewables have seen notable valuation re-ratings. Cameco (CCJ.US), a leading nuclear fuel producer, stands out as a prime example: driven by the renewed mainstream narrative around AI and nuclear power, its shares significantly outperformed the broader market in 2025, becoming a key upstream beneficiary of the “computing power trade.”

The Power Gap Is Widening Rapidly

As AI models continue to scale up, electricity demand from data centers is growing at an exponential pace. Estimates from multiple institutions suggest that incremental data center power demand in the U.S. could reach tens of gigawatts over the coming years—far exceeding the carrying capacity of the existing power grid. Under these conditions, power supply-demand imbalances are no longer a distant risk, but a critical factor shaping real-world decisions, with pressure already reflected in regional power price volatility and the revaluation of power assets.

Structural Contradictions Behind the “Power Shortage”

The challenges facing the U.S. power system stem from a dual squeeze: aging infrastructure and a sudden surge in new load. A large portion of grid equipment is approaching the end of its designed lifespan, while new generation projects and transmission lines require long construction cycles, making it difficult to keep pace with the rapid expansion of AI data centers. At the same time, mismatches among power equipment, grid interconnection capacity, and regional load distribution have caused certain states to encounter power bottlenecks earlier than others, further elevating the strategic value of energy assets.

Big Tech Moves In: Power Grabbing Becomes a New Strategy

In response to potential constraints on computing expansion, leading technology companies have begun to actively intervene on the supply side of energy.

Elon Musk’s xAI has deployed gas-fired generation facilities at data centers to alleviate short-term power shortages. Meta (META.US) has continued to deepen its nuclear energy布局, partnering with companies such as Vistra and Oklo to secure stable baseload power for AI training. Amazon (AMZN.US) has also participated in the development of small modular nuclear reactor (SMR) projects, locking in future energy sources required for computing growth. Google (GOOG.US / GOOGL.US) has gone a step further, directly acquiring wind, solar, and energy storage assets through mergers and acquisitions, fully integrating power security into its long-term strategy.

A Window for Power Industry Re-Rating

In the new AI-driven “power shortage cycle,” electricity is no longer merely a cost item, but a core resource that determines the upper limit of computing expansion. Companies across the power generation chain—including nuclear fuel suppliers, grid equipment manufacturers, and energy storage providers—are being reabsorbed into the long-term AI growth narrative. Market participants widely believe that as AI commercialization enters a phase of tangible earnings delivery, the scarcity of power resources will become increasingly explicit, and energy companies with resource, technological, or regional advantages may continue to benefit from this structural trend.