NVIDIA Vera Rubin confirmed for Q4 2026 shipment — first hyperscaler racks in production by Q1 2027

NVIDIA's Vera Rubin platform — the Blackwell successor unveiled at CES 2026 — is confirmed for Q4 2026 shipment. Hyperscaler procurement teams have rack-scale slots reserved through Q1 2027. NVIDIA also formalized the GTC 2026 LPU (Language Processing Unit) roadmap, slotting three generations of hardware through 2028.

The Q4 Rubin timeline matters because it sets up the head-to-head test against AMD's MI450 Helios shipping in Q3. For ~one quarter, Helios is the only next-gen non-NVIDIA substrate at rack-scale; from Q4 onward, hyperscalers can dual-source on architecture-equivalent generations. That changes pricing leverage on every renewal cycle.

The LPU is the under-discussed part of the roadmap. NVIDIA has historically owned the "general-purpose GPU" lane while custom ASICs (Google TPU, Cerebras WSE, Groq LPU) competed on language-workload specialization. NVIDIA shipping its own LPU collapses that distinction — the GPU vendor entering the language-specialized chip lane is the structural shift, not the raw performance numbers.

Data Center Dynamics — NVIDIA LPU roadmap → · Futurum — NVIDIA Rubin AMD Helios CES →