Dnotitia Brings Dedicated Vector Silicon to Server Scale at AI Infra Summit 2026
New vector silicon just came back from fab. Dnotitia's VDPU could reshape how AI workloads split compute.
Why it matters
A new specialized chip for vector operations at server scale represents a competitive move in the compute buildout — signaling that the Nvidia/AMD duopoly in AI silicon is fragmenting. Vector-specific ASICs could unlock efficiency gains for embeddings and retrieval at scale, a critical path for agentic workloads.
The key facts
5 to knowDnotitia VDPU (Vector Data Processing Unit) first ASIC samples returned from fabrication
Chip-level characterization underway
Announced at AI Infra Summit 2026, Sept 15-17, Santa Clara
Company positioned as AI data infrastructure + semiconductor vendor
Focus on server-scale vector processing (likely embeddings, retrieval-augmented generation)
Go to the source
EnterpriseAIhpcwire.com
Publisher excerpt: SANTA CLARA, Calif., Sept. 18, 2026 — Dnotitia Inc., an AI data infrastructure and semiconductor company, today announced that the first ASIC samples of its Vector Data Processing Unit (VDPU) have returned from fabrication, with chip-level characterization now underway. At AI Infra Summit 2026,…