A research report from East Money Information Co.,Ltd. indicates that the explosion in AI computing power is driving a sharp increase in power demand from data centers in North America. It is projected that the additional power supply demand for new AI data centers in the United States from 2026 to 2030 will reach as high as 125GW. Major manufacturers have order backlogs stretching to 2029-2030, and the expansion cycle for the core component supply chain is lengthy, taking 3-5 years, indicating a persistent hard gap between supply and demand. The investment framework follows the value transmission path from "complete systems/integration → components → materials," focusing on the dual themes of "overseas expansion + domestic substitution." The main views of East Money Information Co.,Ltd. are as follows:
The AI computing power-induced electricity shortage is driving an upswing in the gas turbine market, benefiting supporting companies in the military-industrial "two-engine" supply chain. Gas turbines, as mature power equipment with wide power coverage, rapid start-up capabilities, and stable operation, have long played a key role in providing baseload power and peak shaving for electrical grids. Currently, the world is undergoing a profound transformation triggered by the AI computing boom. The parameters and computational requirements of large AI models are growing exponentially, pushing data centers into the "gigawatt-era." Their characteristics of high energy consumption, high power density, and highly dynamic loads pose significant challenges to existing power grids in regions like the United States. Against the backdrop of slow grid expansion and long construction cycles for transmission infrastructure in the U.S., gas turbines, with their advantages of flexible deployment and reliable power supply, have become the optimal solution for addressing the "power shortage anxiety" of AI data centers in the short to medium term. It is noteworthy that gas turbines and aero-engines share highly overlapping technical pathways, material systems, and supply chains. Therefore, the prosperity in the gas turbine market, driven by surging demand from AI data centers, presents new opportunities for performance growth and a golden period for valuation reassessment for military-industrial supporting enterprises that have long been deeply involved in the "two-engine" field and possess capabilities in special materials development and precision manufacturing.
**Demand Side: AI Computing Power Drives "Rigid" Power Gap, Gas Turbines Become Essential for Overseas Giants**
AI data centers exhibit "three-high" characteristics in electricity consumption, creating a new, rigid demand for power supply. AI data centers are defined by high total energy consumption, high power density, and highly dynamic loads. During the training phase, a single pre-training session for a model like GPT-5 can consume up to 120 GWh of electricity. At the chip level, single GPU power consumption has surpassed 1000W, driving power requirements per rack towards 600kW and beyond. In terms of load, the highly synchronized working patterns of thousands of GPUs cause sharp "sawtooth wave" fluctuations in grid power. This imposes stringent demands on the capacity, density, and response speed of the power supply system that traditional grids struggle to meet, making customer-side gas turbines with rapid start-stop capabilities and stable output a necessity.
The U.S. power grid, constrained by a "triple bottleneck," is unable to handle the load, leading to a severe imbalance between power supply and demand. The U.S. is experiencing a severe shortage of new power generation capacity, with only about 30GW of growth expected by 2030, mostly from intermittent renewable sources that cannot meet the 24/7 stable operation needs of AI data centers. Grid infrastructure is aging, with the average time to build a new transmission line stretching up to 10 years amid complex approval processes. Grid fragility is prominent, with the average outage duration in U.S. data center concentration areas surging over 80% in 2024. According to calculations by East Money Information Co.,Ltd., the additional power supply demand for new AI data centers in the U.S. from 2026 to 2030 is estimated to be as high as 125GW. With limited traditional grid expansion, this creates a vast market space for distributed, customer-side gas turbine power plants.
Customer-side gas turbines emerge as the optimal solution to break the deadlock, with a market potential reaching hundreds of billions. Facing grid bottlenecks, customer-side gas power generation has become the only feasible solution that simultaneously meets requirements for large capacity, high reliability, and rapid deployment. Overseas tech giants are widely adopting this approach; for instance, xAI has purchased five additional 380 MW gas turbines to power its data centers, while projects by Meta and OpenAI also commonly utilize gas turbine solutions. East Money Information Co.,Ltd. estimates that, assuming an average gas turbine price of $300/kW, the market space for new gas turbine installations driven solely by U.S. AI data centers from 2026 to 2030 could reach $22 billion. According to GasTurbineWorld estimates, the global after-sales service market for gas turbines is expected to reach $44.5 billion in 2025 and will continue to grow with the operation of new installations, with maintenance and repair demand for aftermarket products and services projected to exceed $261 billion from 2025 to 2034.
**Supply Side: Global Giants Ramp Up Production After Selling Out Capacity, Domestic Core Suppliers Face Historic Entry Window**
Orders for the three major overseas giants are overflowing, with capacity sold out through 2028. The international gas turbine market is dominated by three giants: GE Vernova, Siemens Energy, and Mitsubishi Heavy Industries, which collectively hold an 85% market share. Driven by demand from AI data centers, orders for these giants have surged. GE Vernova has sold out all its capacity for 2028, with only 10% remaining for 2029. Siemens Energy reported a 63% year-on-year increase in gas turbine orders for fiscal year 2025, with delivery schedules extending to 2030. Mitsubishi Heavy Industries plans to double its capacity within the next two years. Despite expansion efforts, constrained by complex supply chains and key component bottlenecks, the global market exhibits severe "seller's market" characteristics.
The core component supply chain is highly concentrated, with expansion constrained by "rigid bottlenecks." Core hot-section components of gas turbines possess high technical barriers, and the global market has long been monopolized by a few suppliers like PCC and Howmet. With the recovery of the aviation industry prioritizing engine demand, the core component supply chain has become the biggest "rigid bottleneck" for gas turbine capacity expansion. This forces overseas system integrators to actively seek reliable alternative capacity to ensure delivery capabilities. The severe supply-demand mismatch shifts pricing power to sellers, and profits across the industrial chain are expected to increase significantly.
Domestic "two-engine" suppliers are poised to seize a historic entry window, with potential for a significant jump in their supporting share. Domestic core component suppliers for the "two-engine" sector, represented by companies such as Yingliu Co.,Ltd., Wanze Co.,Ltd., and HangYu Technology, have, after years of deep cultivation, passed stringent certifications from international leaders like GE, Siemens, and Baker Hughes. They have established solid international competitiveness in areas like blades and disc/ring components. Against the backdrop of global supply chain tightness, leveraging their technical prowess, cost advantages, and flexible capacity response, domestic "two-engine" suppliers are expected to secure long-term contract orders transferred from overseas giants. This could facilitate an upgrade from their role as "domestic suppliers" to becoming "global core suppliers," potentially allowing them to deeply benefit from this high-growth cycle.
**Investment Recommendations**
Gas turbines are technology-intensive high-end equipment, often hailed as the "jewel in the crown" of machinery. Core hot-section components must operate under extreme conditions exceeding 1400°C, representing high technical barriers. The gas turbine industrial chain features a three-tier structure: "materials, components, complete systems/integration." The global market for complete systems is oligopolistic, dominated by GEV, Siemens, and Mitsubishi. Components and materials, however, share technological origins and manufacturing processes with aero-engines, giving domestic "two-engine" industrial chain companies inherent advantages for technology and capacity migration.
The explosion in AI computing power is driving a sharp increase in power demand from data centers in North America. It is projected that the additional power supply demand for new AI data centers in the United States from 2026 to 2030 will reach as high as 125GW. Major manufacturers have order backlogs stretching to 2029-2030, and the expansion cycle for the core component supply chain is lengthy, taking 3-5 years, indicating a persistent hard gap between supply and demand. The investment framework follows the value transmission path from "complete systems/integration → components → materials," focusing on the dual themes of "overseas expansion + domestic substitution."
Regarding specific investment targets, it is suggested to focus on: * System Integrators & Complete System Manufacturers: Jereh Co.,Ltd., Dongfang Electric Corporation Limited, Shanghai Electric Group Company Limited, Qilun Technology Co.,Ltd.. * Core Component Suppliers: Yingliu Co.,Ltd., Wanze Co.,Ltd., AECC Aviation Power Co.,Ltd., HangYu Technology, Himile Mechanical Science and Technology(Shandong)Co.,Ltd., L&B Co.,Ltd., Hangya Technology Co.,Ltd.. * Core Material Suppliers: Lontrue Co.,Ltd., Shangda Co.,Ltd., AECC Aviation Materials Co.,Ltd..
**Risk Warnings**
* Market Demand Fluctuation Risk: The construction progress and capital expenditure of overseas AI data centers are uncertain. If tech giants slow investments or shift technological routes, it could directly impact gas turbine demand. * International Trade and Policy Risk: Geopolitical tensions, trade barriers, or changes in export control policies could significantly disrupt companies' order acquisition, production delivery, and supply chains. * Exchange Rate Fluctuation Risk: Companies' overseas orders are mostly settled in foreign currencies like USD. Significant fluctuations in the RMB exchange rate will generate exchange gains or losses, potentially directly affecting financial performance reported in local currency.
Comments