Anthropic and OpenAI Are Both Racing to IPO. GE Vernova Wins Either Way.

GE Vernova's Q2 2026 orders surged 88% year over year to $24.2 billion, while electrification orders from data centers exceeded $5 billion in H1 2026—more than double its total data-center orders for all of 2025. The company is positioned to benefit from rising power demand tied to OpenAI and Anthropic, which are pursuing IPOs and require substantial new electricity capacity for AI infrastructure. Shares remain up roughly 45% year to date despite pulling back from June highs, though the stock trades at an elevated approximately 38x forward earnings.

Analysis

GEV's upside is less about incremental AI capex headlines than the duration and pricing power embedded in grid and generation bottlenecks. Gas-turbine lead times, transformer scarcity, and transmission interconnection delays allow suppliers with installed-base service relationships to convert equipment demand into higher-margin multiyear service revenue; ETN, HUBB, PWR, and MYRG are cleaner second-order beneficiaries of the electrical-balance-of-plant constraint. The risk is that GEV's valuation already capitalizes a sustained order-to-revenue conversion cycle, leaving little tolerance for project delays or lower-margin mix.

Over the next 1-3 months, the key catalyst is evidence that large-load data-center contracts are translating into funded utility procurement rather than nonbinding capacity reservations. A stronger read-through would be rising backlog conversion, improved Power segment margins, and continued pricing on outage/service agreements; these matter more than AI-company financing announcements. The 6-18 month structural winner may be regulated utilities and merchant generators with deliverable capacity near data-center hubs—CEG, VST, and selected utilities—not equipment vendors, if power scarcity enables rate-base expansion or premium bilateral power contracts.

The contrarian view is that the market is conflating announced data-center demand with near-term turbine shipments. Permitting, interconnection, gas-pipeline availability, and local political resistance can defer projects by years, while customer-funded electricity costs may reduce the urgency for uneconomic model-training expansion. A material downward revision to backlog conversion, Power margin guidance, or an easing in turbine lead times would likely drive disproportionate multiple compression in GEV versus more diversified electrical suppliers.

Key Decisions for Investors

• Do not chase GEV solely on AI-IPO headlines; accumulate only on a 10-15% pullback or after earnings confirm backlog conversion and segment-margin expansion. Thesis horizon: 6-18 months; exit if management cuts full-year Power margin or free-cash-flow guidance.
• Express the grid bottleneck through a diversified pair: long ETN and HUBB versus short GEV in equal dollar amounts for the next 3-6 months. ETN/HUBB have less exposure to a single turbine-cycle normalization and greater direct exposure to distribution, protection, and data-center electrical equipment; close if GEV reports materially stronger service-margin acceleration than peers.
• Add CEG or VST on confirmation of new long-duration data-center power contracts or capacity-price strength, rather than on equipment-order news. Target 6-18 months; principal risks are power-price retracement, adverse state regulation, and accelerated new generation supply.
• Set a diligence alert around utility capital plans and GEV quarterly disclosure: require evidence that data-center-related orders are backed by transmission/interconnection approvals and deposits. If cancellations or deferred delivery schedules emerge, reduce GEV exposure before consensus revenue estimates reset.