# RISC-V HPC leaders chart path from skepticism to supercomputing deployment at SC26

> **Key takeaway:** RISC-V has transitioned from academic curiosity to active supercomputing deployment, with server-grade silicon tapeouts, ratified specifications, and commercial accelerators like InspireSemi's Thunderbird now entering the market.

**Published:** 2026-11-17T11:40:16+00:00 
**Source:** RISC-V International 
**Category:** open-hardware 
**Canonical URL:** https://fosswire.org/news/risc-v-hpc-leaders-chart-path-from-skepticism-to-supercomput.html 

## Summary
At the SC26 supercomputing conference in Chicago, RISC-V International's HPC Special Interest Group leaders Doug Norton and Dr. Nick Brown detailed how the architecture has moved from simulated exploration to practical supercomputing deployment within three to four years. Nick Brown, who chairs the HPC SIG and leads the RISC-V testbed at EPCC, noted that early 2023 presentations relied on simulators and basic embedded boards, whereas the field has now produced over 50 published papers and gained access to increasingly powerful multi-core processors and accelerators. Doug Norton, vice-chair of the SIG and CMO of InspireSemi, described how industry analyst predictions of a 5-10 year timeline have collapsed, with RISC-V now 'happening globally, out in the open' in 2026.

The convergence of AI and traditional HPC workloads is accelerating RISC-V adoption at the highest levels of computing. Multiple vendors have taped out server-grade silicon built on the RVA23 profile, and the Server Platform specification has been ratified. InspireSemi's Thunderbird RISC-V accelerator for HPC and AI is now entering commercial deployment through partnerships with Lenovo, Dell Technologies, Penguin Solutions, and E4 Computer Engineering, targeting both PCIe accelerator cards and integrated HPC servers. The EuroHPC-funded DARE project is also developing RISC-V technologies for future European supercomputers, while Brown's team at EPCC operates alongside ARCHER2, the UK's national supercomputer.

## RISC-V supercomputing ecosystem developments
- **HPC-AI convergence:** The distinction between traditional HPC (FP64 scientific computing) and AI workloads is blurring, with organizations adopting terms like 'Advanced Computing' or 'AI supercomputing'; RISC-V accelerators designed for AI are proving effective for traditional HPC workloads.
- **Matrix extensions roadmap:** The RISC-V AME (Advanced Matrix Extension), IME, and VME matrix extensions are identified as key upcoming milestones that will further strengthen RISC-V's position in high-performance computing.
- **Commercial deployment pipeline:** InspireSemi's Thunderbird accelerator is moving to market through established server OEMs and HPC integrators, with deployment options including both PCIe add-in cards and full HPC server configurations.
- **European sovereign computing:** The EuroHPC Joint Undertaking's DARE project is actively funding RISC-V technology development specifically for future European supercomputers, parallel to commercial efforts.

## Why it matters
For Linux distributions and HPC system administrators, the maturation of RISC-V server platforms means new hardware options that require kernel and toolchain support for RVA23-compliant silicon and emerging matrix extensions. The entry of major OEMs like Lenovo and Dell into RISC-V accelerator distribution signals that software ecosystem investment—particularly in compilers, MPI implementations, and math libraries—will need to accelerate to match hardware availability. For open hardware advocates, the combination of open ISA specifications, sovereign European funding through DARE, and commercially competitive silicon represents a credible alternative to proprietary architectures in a market segment historically dominated by x86_64 and ARM.