AI Capex Surge Sends Power Equipment Stocks Soaring as Grid Investment Lags 70% Behind Generation

AI Capex Surge Sends Power Equipment Stocks Soaring as Grid Investment Lags 70% Behind Generation A $2.4 billion gas-fired power contract and a $12 billion pipeline reveal how AI demand is reshaping energy infrastructure investment. Babcock & Wilcox landed a $2.4 billion design-build contract with Base Electron for 1.2 gigawatts of natural gas-fired power generation, driving the company's backlog up 470% to $2.8 billion. Shares closed at $14.54, up 129% year to date. Base Electron is evaluating another 1.2 GW option, and Babcock's global pipeline exceeds $12 billion. The numbers are extraordinary for a company that is not yet profitable and carries negative stockholders' equity of $131.5 million.¹ That contract matters because it is physical proof of what the AI spending boom looks like when it hits the energy sector. Hyperscalers like Amazon and Meta are racing to secure power from nuclear, natural gas, solar, and every other available source. Data centres already account for more than 1% of global electricity use, according to the IEA. The question is no longer whether AI will drive power demand growth. It is whether the grid can be built fast enough to deliver it.^4,7 The IEA's World Energy Outlook flagged the gap directly. Investments in electricity generation have surged nearly 70% since 2015, but spending on power grids has increased at less than half that rate. That mismatch creates bottlenecks. New generation capacity is useless if it cannot connect to load centres, and data centres are appearing in locations driven by fibre connectivity and land availability, not grid capacity.⁶ Wood Mackenzie framed the challenge as geopolitical. The global power sector is undergoing transformational change, with demand growth set to accelerate from emerging market development, advancing electrification, and the AI-driven data centre boom. But whether supply can be ramped fast enough is an open question. The firm's inaugural energy forum examined this tension directly.⁸ Montel's Huangluolun Zhou examined what the surge means for European grids specifically. The AI boom is driving rapid, localised growth in data centre demand that European grid operators and policymakers are not prepared for. The demand is concentrated, arriving in specific locations, and it needs firm baseload power, not intermittent renewables. That profile favours gas-fired and nuclear generation.³ The nuclear thesis is already running. Uranium Energy Corp offers concentrated exposure to AI's nuclear baseload demand, and Fluence Energy ran 98% in a single week on AI power optimism. Management at Babcock guided 2026 core adjusted EBITDA to $70 million to $85 million, roughly 80% year-on-year growth, excluding any data centre upside. The market is pricing in AI power demand before the megawatts are connected.¹ Big tech's capital expenditure tells the story from the demand side. Capex across the hyperscalers is in the hundreds of billions and could test $1 trillion in the next year. Investors are starting to worry about a potential bubble. They want to see returns, not just spending announcements. But the physical infrastructure being contracted suggests the spending is converting into real power demand, not just GPU purchases.⁵ The IEA has warned that AI's energy appetite is outpacing deployment of AI-based climate solutions. The agency found that the deployment of AI tools to curb energy intensity is not keeping pace with the sector's voracious demand for power. The irony is uncomfortable: a technology promoted as a climate solution is, in its current form, accelerating fossil fuel consumption.⁹ Europe's energy crisis adds a second dimension. Just four years after Russia's invasion of Ukraine, the Middle East conflict once again underscores the limitations of relying on gas imports and storage, Montel reported. European policymakers are rethinking gas storage requirements at the same time that data centre operators are requesting grid connections for hundreds of megawatts of firm power. The two demands compete for the same molecules and the same infrastructure.² The risk for Babcock sits in its balance sheet. Negative stockholders' equity and a 6.50% note refinancing due in 2026 mean the company needs to execute on its backlog without stumbling on capital structure. A $12 billion pipeline is meaningless if the company cannot finance the working capital to deliver it.¹ The signal to watch is whether Base Electron exercises its option for the second 1.2 GW tranche. If it does, the AI-to-gas-turbine pipeline doubles for a single counterparty, and the market will need to reprice the speed at which AI demand is translating into physical power infrastructure orders.¹