DEPLOY

Foundry · Company record

SMIC

3 AI chips are made at SMIC, designed by 1 different company.

Process-node breakdown

SMIC's AI chip work, grouped by which manufacturing process was used. Chips whose process isn't public show up under "unspecified".

  • SMIC N+2 (~7nm)3 chips

Fastest chips SMIC makes (top 3)

Peak compute per chip, straight from each vendor's datasheet (FP8 where published, otherwise FP16/BF16).

  1. 12,400 TFLOPS
  2. 2800 TFLOPS
  3. 3400 TFLOPS

All chips SMIC makes, by designer

Huawei3 chips

  • Huawei Ascend 910BSMIC N+2 (~7nm)

    Huawei's flagship domestic AI training accelerator, positioned as the primary Chinese alternative to Nvidia H100 after US export controls. Uses HBM3, Da Vinci architecture. Widely deployed at Chinese hyperscalers and government cloud regions.

  • Huawei Ascend 910CSMIC N+2 (~7nm)

    Follow-on to 910B (2024), targeting ~60% of H100 training throughput on Da Vinci 3.0. Volume production ramping despite HBM supply constraints from Samsung/SK Hynix export restrictions.

  • Huawei Ascend 910DSMIC N+2 (~7nm)

    Announced 2025 as the next generation targeting Nvidia B200 territory. Chiplet architecture per public disclosures; volume production timeline uncertain given SMIC 5nm-equivalent process constraints.