Company comparison
Axelera AI vs FuriosaAI
Radar profile, momentum bars, and stack placement side by side.
Axelera AI
Not scored
momentum
FuriosaAI
Not scored
momentum
Axelera AI capital
—
FuriosaAI capital
—
No radar series.
Momentum head-to-head
Attribute tape
| Field | Axelera AI | FuriosaAI |
|---|---|---|
| Category | Semiconductors and hardware | Semiconductors and hardware |
| Stack | Layer 1 | Layer 1 |
| HQ | Eindhoven, Netherlands | Seoul, South Korea |
| Founded | 2021 | 2017 |
| Status | Operating | Operating |
| Funding | — | — |
| Momentum | — | — |
| Summary | Axelera AI is a European AI semiconductor company designing hardware and software for efficient edge AI inference. Its Metis platform combines proprietary Digital In-Memory Compute (D-IMC) architecture with RISC-V dataflow cores to deliver high-throughput, low-power AI acceleration for computer vision and generative AI workloads at the edge. | FuriosaAI is a South Korean fabless semiconductor company that designs AI inference accelerators (NPUs) for data-center workloads, including large language models and computer vision. Its hardware-software co-design approach — anchored by the proprietary Tensor Contraction Processor (TCP) architecture and RNGD accelerator family — targets high performance per watt and per dollar as an alternative to GPU-centric AI infrastructure. |
| Who for | Teams evaluating AI vendors in this category. | Teams evaluating AI vendors in this category. |
| Differentiator | Axelera AI is a European AI semiconductor company designing hardware and software for efficient edge AI inference. Its Metis platform combines proprietary Digital In-Memory Compute (D-IMC) architecture with RISC-V dataflow cores to deliver high-throughput, low-power AI acceleration for computer vision and generative AI workloads at the edge. | FuriosaAI is a South Korean fabless semiconductor company that designs AI inference accelerators (NPUs) for data-center workloads, including large language models and computer vision. Its hardware-software co-design approach — anchored by the proprietary Tensor Contraction Processor (TCP) architecture and RNGD accelerator family — targets high performance per watt and per dollar as an alternative to GPU-centric AI infrastructure. |
