AI is reshaping every layer of the compute stack — from data-centre training clusters down to milliwatt inference engines at the edge. The hardware powering it is overwhelmingly proprietary: a handful of architectures, gated by licensing terms that dictate who is allowed to build, adapt, and deploy AI systems.
RISC-V is a different path: an open, extensible instruction-set architecture that lets nations, companies, and research labs build sovereign AI infrastructure on their own terms — from AI Factories in the cloud to embedded inference in the field.
GROW-AI is the second edition: after São Paulo in 2025, and inheriting the Sino-European Workshop lineage (Tirana 2024, Hong Kong 2024).
Hosted at TU Wien, GROW-AI 2026 convenes researchers, industry, and policy-makers from Europe, Asia, and Latin America. Three days, three frames: open technology strategy on Monday; AI-native silicon on Tuesday; open-source AI toolchains on Wednesday — from the policy that lets open hardware exist to the compilers that make it usable.
Venue
GROW-AI 2026 is held at TU Wien, in TUtheSky — the rooftop venue at the top of the BA building. TU Wien is Austria’s largest technical university, in Vienna’s 4th district, a few minutes from the city centre on foot or by U-Bahn.
Why now
Open silicon for AI is a strategic question, not a technical one. The technology exists — the vector extension is ratified and upstream, matrix extensions are converging, the toolchains compile, the silicon ships. What’s missing is alignment: shared roadmaps across regions, shared profiles across vendors, and fabrication pathways that don’t end in a single jurisdiction’s export-control list. GROW-AI 2026 is where that alignment is negotiated. Think global, act local. Pool resources on shared interoperability. Differentiate at the implementation layer.