Breaking

Soitec Teams With ZenSemi to Scale 300mm BCD-on-SOI

June 29, 2026 at 06:24 EDT

Infra & Chips
Robotics

French semiconductor-materials maker Soitec and Chinese specialty foundry ZenSemi said on June 29, 2026 that they had formed a strategic partnership covering advanced power electronics for AI data centers, humanoid robots, electric vehicles (EVs) and industrial equipment. Under the arrangement, Soitec will supply its 300mm Power-SOI substrates while ZenSemi develops and scales a 300mm BCD-on-SOI process, with both companies framing the deal as a way to enable high-volume production and provide a high-performance manufacturing platform for fabless companies and IDMs. Financial terms were not disclosed.

June 29, 2026 · Soitec × ZenSemi

France meets China to mass-produce power chips for the AI era

Soitec will supply its 300mm Power-SOI substrates to Guangzhou foundry ZenSemi to scale a new BCD-on-SOI process — targeting AI data centers, humanoid robots, EVs and industrial power electronics.

~30%

smaller die vs. conventional bulk BCD, in first-silicon validation

300mm

Power-SOI wafer scale (12-inch) targeted for high volume

+5.8–8%

Soitec shares after the news, to around €120

Die size: BCD-on-SOI vs. conventional bulk BCD

Same function, smaller silicon — taller column = larger die area

100%

Conventional
bulk BCD

~70%

BCD-on-SOI

Why SOI instead of bulk silicon

• Dielectric isolation → no latch-up

• Reduced parasitic effects

• Better EMI / EMC resistance

• Low leakage at high temperature

• Easy high-density logic + high-voltage integration

The supply flow

Soitec

300mm Power-SOI substrates

→

ZenSemi

BCD-on-SOI process + capacity

→

Power ICs

AI · robots · EVs · industry

Promising signal

First-silicon validation already shows a ~30% die-size cut, and pairing SOI with a Chinese foundry at 300mm scale is a notable industry first.

Still early

Financial terms, timeline, pricing and concrete benchmarks are undisclosed. No production products or user cases exist yet — it sits at process development and ramp.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.