DigiTimes：AI数据中心电力紧缺致全球燃气/SOFC发电设备交期拉长至2-5年，台厂熙特爾子公司歐羅仕聯手Volvo Penta代理預訂300台燃氣機組(約120MW)搶攻2027快速交付

The focus of the global AI race has shifted from securing advanced chips to securing access to electricity, with electrical power now becoming the deciding factor for computing capacity. As power consumption surges at AI data centers (AIDCs), conventional public grids are increasingly facing bottlenecks, and self-generation and improved grid resilience are now requirements in major data center construction projects worldwide.

Due to the rapid emergence of AIDCs and their massive power needs, as well as long-term ESG and emissions-reduction concerns, supply chain sources point out that data centers worldwide are shifting from traditional diesel generators to lower-carbon liquefied natural gas (LNG) gas-fired generator sets for their self-generation needs. However, both conventional gas-fired generator sets and new hydrogen fuel-cell systems are currently facing long order backlogs worldwide.
As a result of this global production bottleneck, power-generation equipment delivery times have now lengthened to two to three years after customers place orders. Wait times at some large-scale projects have even reached as long as five years, leading to significant delays in the deployment of computing infrastructure. Major German equipment manufacturers, leading US solid oxide fuel cell (SOFC) solution provider Bloom Energy, and international power-generation heavyweights such as GE and Mitsubishi Heavy Industries (MHI) are all facing this capacity crunch.
To address these delays, Aurosi Precision—a subsidiary of Taiwan-based Seetel New Energy—recently launched a strategic partnership with Jouco Group, the Taiwan distributor for Swedish power-equipment giant Volvo Penta.
The partners have secured production capacity in advance, and have arranged to supply 300 gas-generator engines through 2027. The engines will provide an estimated 120MW of generating capacity, with localized modular production at the center of their strategy to capture fast-delivery opportunities in the computing power market.
In contrast to the complex construction processes used at traditional large-scale power plants, the partnering companies will adopt a 10MW-class containerized modular design, with generator sets integrated directly into 40-foot containers, which will facilitate on-site lifting and rapid cabling to significantly shorten construction timelines.
More importantly, while international manufacturers often require several years to deliver equipment, deliveries of this modular architecture can begin as early as 2027. The partners intend to leverage this advantage in rapid deployment to secure a position in the market as an emergency response power provider for AIDCs.
In terms of operating costs and system architecture, the gas-fired generator sets cost about one-third as much to procure as SOFC hydrogen equipment. Moreover, the 500–600℃ waste heat generated during operation can be recovered to restore condensate and provide hot water, and the system can also be used to support AI server cooling requirements, in a boost to overall efficiency.
Time-to-market is central to AIDC construction, and the speed at which computing capacity comes online has a direct impact on corporate competitiveness. With major global power-generation equipment makers now facing capacity constraints, and delivery times now ranging in years, hybrid systems combining gas-fired generation and energy storage that offer local assembly, flexible containerized deployment, and rapid delivery have become a crucial solution to the global power bottleneck in the AI race.