Перейти к содержимому
Новости

NextGO Epi raises €2 million pre-seed round to advance gallium-oxide power-device supply chain

Редакция ЦОД:Главное

TechNews 科技新報 — иллюстрация к материалу
Источник изображения: TechNews 科技新報

German startup NextGO Epi completed a €2 million pre-seed round in July 2026 to develop gallium-oxide epitaxial wafers for power electronics. The company identifies electric vehicles and AI data centers as prospective application markets, but does not yet supply finished devices or report deployments in operating facilities.

Главное
  • NextGO Epi closed a €2 million pre-seed round in July 2026.
  • The company supplies gallium-oxide epitaxial wafers, not finished power devices.
  • Finished devices currently require contract manufacturing in Germany, the United States and Taiwan.
  • AI data centers are a target market, but no operating data-center deployments, customers or shipment volumes were reported.
  • Potential power-efficiency benefits remain company claims or theoretical estimates and require validation in real systems.
  • Founder expectations for adoption within two to three years and foundry order-taking by 2028 are not confirmed market commitments.

NextGO Epi, a German startup founded in 2025, has closed a €2 million pre-seed financing round, completed in July 2026. The company develops gallium-oxide epitaxial wafers for power-electronics applications and is targeting eventual use of devices based on the material in electric vehicles and AI data centers.

The company is a wafer supplier rather than a manufacturer of finished power semiconductors. To produce devices, it currently funds contract manufacturing arrangements with fabs in Germany, the United States and Taiwan. NextGO Epi says it is the only European supplier capable of industrial-scale production of 4-inch, industrial-quality gallium-oxide epitaxial wafers; this is a company claim.

For data-center infrastructure, the potential relevance is at the power-conversion layer. NextGO Epi argues that gallium oxide could reduce voltage-conversion losses and associated heat generation. It cites a 4.9 eV bandgap and states that, for the same voltage rating, the required layer thickness can be one-half to one-third that of silicon carbide. The company also cites a theoretical estimate that device resistance could be one-fifth to one-tenth of comparable silicon-carbide and gallium-nitride devices. These are technical claims and theoretical estimates, rather than independently verified operating results in data-center power systems.

Founder and CEO Dashun Zhou expects gallium-oxide components to enter electric-vehicle and AI-data-center applications within two to three years. He further forecasts that fabs could begin independently accepting orders for gallium-oxide components by 2028. Both points are expectations rather than confirmed market commitments or production milestones.

Commercial scaling will depend on a broader device-manufacturing supply chain. NextGO Epi remains reliant on external fabs for conversion of its wafers into components, while major manufacturers cited in the source do not yet have production lines for gallium-oxide devices. The material also faces technical constraints noted by the company, including effectively absent p-type doping and lower thermal conductivity than silicon carbide.

The development therefore represents an early-stage materials and supply-chain initiative, not a verified deployment in an operating AI data center. If device performance and manufacturability are validated, gallium oxide could provide an additional option for selected high-voltage power-conversion stages in future AI infrastructure.

Почему это важно

The financing is an early signal of efforts to introduce gallium oxide into the AI data-center power-conversion chain. If its material and manufacturing ecosystem reach industrial maturity, gallium oxide could become an alternative to silicon carbide and gallium nitride in selected high-voltage conversion stages; that outcome remains dependent on foundry availability and validation in real power systems.

Источники

Материал подготовлен редакцией ЦОД: Главное на основе указанных первоисточников.