The AI Power Trade Has Moved From Chips to the Boilermakers — Babcock & Wilcox Is Up 129% on Gas-Fired Baseload

The AI Power Trade Has Moved From Chips to the Boilermakers — Babcock & Wilcox Is Up 129% on Gas-Fired Baseload An old-line power-equipment maker's backlog jumped 470% on a single data-center contract, the clearest sign that AI's near-term winners are the firms that build and power the buildings, not the ones inside them. The AI power trade has found its way to the unglamorous end of the energy sector. Babcock & Wilcox, a maker of industrial power-generation equipment, has pivoted into AI data-center baseload, and its shares closed at $14.54, up 129.34% year to date.¹ The catalyst was a single deal: a $2.4 billion design-build contract with Base Electron for 1.2 gigawatts of natural gas-fired power, which drove the company's backlog up 470% to $2.8 billion.¹ When a legacy boilermaker's order book more than quintuples on one data-center contract, the AI buildout has reached the picks-and-shovels layer. It matters because it shows where the near-term value is, and it is not in the chips. The companies that actually build and power data centers, the equipment makers, the utilities, the engineering firms, are capturing the spending that the AI boom requires before a single GPU runs. Management guided 2026 core adjusted EBITDA to $70 million to $85 million, roughly 80% year-on-year growth, and that excludes further data-center upside.¹ Base Electron is evaluating another 1.2 GW option, and the company's global pipeline exceeds $12 billion, which is the scale of demand pulling these industrial names into a value re-rating.¹ The fuel choice in that contract is the part the market should not miss. The 1.2 GW Babcock is building is gas-fired, not solar or nuclear, which confirms that the near-term answer to data-center power demand is conventional generation.¹ The data-centre expansion required to support AI is expected to keep fossil fuels in use for longer, according to BloombergNEF, and a flagship contract for gas-fired baseload is exactly what that thesis looks like in practice.³ The value trade in industrials is, at its core, a bet that gas turbines get built before clean baseload can scale. The same demand is reshaping the utilities, which is the other half of the value trade. NextEra's roughly $67 billion acquisition of Dominion, the largest power-utility deal on record, was tied directly to data-center-driven load growth.⁵ Utilities that were treated as sleepy yield plays are being repriced as the infrastructure that delivers AI's power, and the consolidation wave reflects investors paying up for scale in a sector suddenly central to the technology boom. The economics can even help ordinary customers, which strengthens the bull case for the utilities. PG&E estimates that adding a gigawatt of load could lower bills by up to 2%, because a large new customer spreads the utility's fixed costs across more demand.⁴ A data center that pays its share is a benefit to the rate base, not a burden, and that reframes the utilities as beneficiaries rather than victims of the load surge. Hyperscalers are also securing their own supply, with Alphabet paying $5 billion for the solar-and-storage developer Intersect Power.⁴ The constraint that could cap the trade is the grid itself, and it is most visible in the emerging markets. In Southeast Asia, data centers, EVs and green industrial parks are forecast to add more than 100 terawatt-hours of demand and require over $200 billion of investment, yet the report flagging that growth also points to an estimated $18 billion annual shortfall in grid investment by 2035.² If the wires cannot be built fast enough, the demand that is driving the value trade gets stranded, and the equipment and utility names that depend on it stall with it. The risk in the individual names is balance-sheet, not demand. Babcock's legacy balance sheet carries stockholders' equity of negative $131.5 million and a 6.50% note refinancing due in 2026, the kind of financial fragility that a 129% rally can paper over but not erase.¹ A turnaround riding a data-center wave still has to refinance its debt, and that is the catch in the value-trade story. The signal to watch is whether the order pipeline converts to delivered EBITDA and whether the grid keeps pace.^1,2 If Babcock turns its $12 billion pipeline into earnings and utilities keep consolidating around data-center load, the industrials value trade has legs. If grid bottlenecks stall the buildout or the balance sheets crack, the re-rating reverses. For now, the AI power trade is being won by the firms that build the plant, not the ones that fill it with chips.^1,5