GE Vernova
Gas turbines, grid equipment, and nuclear — the dispatchable-power bottleneck assets of the AI data center buildout. Q2 2026 (July 22): revenue $11.1B, orders $24.2B (+88% YoY), record backlog $176B; gas-turbine backlog reached 116 GW (from 100 GW in Q1, guided to ≥125 GW by year-end). The Q2 tell: for the first time, named frontier AI labs — not just hyperscaler cloud arms — are contracting multi-GW power directly (Microsoft's 7-turbine, 2.7 GW Texas data center; Meta, Google, OpenAI and Anthropic reportedly queuing), and electrification backlog hit $42B with more than $5B of data-center orders year-to-date. Adjusted EPS ($2.47) missed the $3.04 estimate — a margin-execution flag the score does not yet reflect.
Scenarios
Sold-out gas turbine slots into 2030 in a three-player oligopoly (GEV, Siemens Energy, Mitsubishi) equal multi-year pricing power — every AI data center wave needs dispatchable generation plus grid equipment. SMR optionality is real — Darlington BWRX-300 under construction, up to $40B US-Japan government support for Tennessee/Alabama builds.
Revenue recognition is 2027-2030 weighted — if AI capex peaks before slots convert, order momentum reverses while capacity expands into a glut (20 to 24 GW/yr output target by 2028). Wind segment remains a profitability drag, and the stock prices in continued order beats.
Key Factors to Watch
- ●Q1 2026: orders $18.3B (+71% organic), backlog $163B, $200B expected in 2027; 2026 guidance raised
- ●Gas backlog + slot reservations at 100 GW (83 GW prior quarter), sold out into 2030 at improving prices; data centers ~20%
- ●Electrification orders +111% YoY — $2.4B data-center equipment orders in Q1 alone, more than all of 2025
- ●SMR/nuclear optionality — Darlington BWRX-300 base mat approved; up to $40B US-Japan support framework
- ●Revenue lags orders (2027-2030 weighted) — AI-capex cycle timing vs capacity expansion is the key risk
Score History
| Date | Score | Direction | Note |
|---|---|---|---|
| 2026-08-16 | 7.8 | Positive | Score 7.55→7.8 (are 7→8). Q2 2026 (July 22) marked the first time named frontier AI labs — not just cloud arms — contracted multi-GW power directly: Microsoft's 2.7 GW Texas data center (7 turbines), with Meta, Google, OpenAI and Anthropic reportedly queuing; electrification backlog $42B with >$5B of data-center orders YTD; gas backlog 116 GW. AI revenue exposure broadened from generic hyperscaler demand to direct frontier-lab contracts. The EPS miss ($2.47 vs $3.04) is a margin/execution watch — held on moat_durability. |
| 2026-06-11 | 7.5 | Positive | Initial assessment — Ondra portfolio holding, added on BF coverage-gap signal (inbox 2026-06-10). Score 7.55 = AI power demand drives record orders and a sold-out gas oligopoly into 2030 (md 8, dr 2), but revenue exposure is broad energy, not AI pure-play (are 7) — slots between Vertiv 7.6 and NextEra 7.5. Q1 2026 orders +71% organic, backlog $163B, guidance raised. |
Manufacturing Peers
Last researched: 2026-08-16
This is research and analysis, not financial advice. Scores reflect AI impact potential, not investment recommendations.