Fractile

Compute-in-memory (CIM) AI memory chips and inference racks that let data centers support more users and longer context at the same power.

In-house AI inference chips and full-stack computing hardware/software for high-density AI data centers

AI inference chips and systems for data centers

Series C · $450M · 2025-11 · Valuation $2.0B

3D compute-in-memory (3D-CIM) AI inference chips

Series C · $293M · 2026-10

3D compute-in-memory AI inference chips: 64 TOPS CIM units with in-house RISC-V vector architecture, 6GB-48GB 3D DRAM

Angel · $15M · 2026-10

Neuromorphic sensing-memory-compute chips, using brain-inspired computing and mixed-signal design to integrate sensing, storage and computation

Series C · $139M · 2026-10

Photonic AI inference chip (DX-1/optical tensor processing unit) that uses SRAM plus photonic interconnects to bypass the HBM bottleneck, with accompanying optical networking and rack-scale systems

Series B · $532M · 2026-08 · Valuation $3.3B

High-performance AI inference chips for automotive, mobile and data center applications

Undisclosed · $46M · 2025-05

Full-stack AI cloud platform: GPU cloud compute, AI training/inference infrastructure and integrated data center plus power services

Debt / Convertible · $3.7B · 2026-09 · Valuation $15B

Designs the AI inference chip Rebel100 along with companion servers and full racks, for telecom operators and data centers to run agent inference

Pre-IPO · $898M · 2026-03 · Valuation $2.3B

Develops the RNGD data center inference accelerator card and the WARBOY vision NPU for enterprises running large model inference

Series D+ · $745M · 2026-10 · Valuation $2.9B

MEMS-based silicon clock oscillator chips replacing quartz devices, providing precision timing for AI data centers, consumer electronics and industrial systems.

Series B · $55M · 2026-06