Ax3.ai
Home / News / The Ax3 take
ReutersNVIDIA2h ago

NVIDIA reveals Vera Rubin details — general availability slips to H2 2027

The Vera Rubin NVL72 platform, positioned as the successor to Blackwell, will ship six months later than previously indicated at GTC. Reservations open in Q3 2026.

NVIDIA has published detailed specifications for its Vera Rubin NVL72 rack-scale platform while confirming that general availability moves to the second half of 2027 — roughly six months later than the timeline shown at GTC.

The company attributed the shift to a broader packaging transition and to prioritizing yield on the Vera CPU, which anchors the platform's 72-GPU coherent domain. Reservation books open in Q3 2026, with early allocation weighted toward existing GB300 customers.

Cloud providers contacted for comment said the slip was expected internally and is already reflected in 2027 capacity planning. Several noted that Blackwell-generation supply remains the binding constraint through at least mid-2027.

The Ax3 take

What it means if you're buying: Blackwell holds its value longer. A six-month Rubin slip extends the useful pricing life of B300 and GB300 clusters — capacity you can secure today keeps its competitive position deeper into 2027.

What to do about it: If you were holding out for Rubin, lock Blackwell now while lead times are measured in weeks, not quarters. On the Ax3 marketplace, SIRIUS-2048 (2,048× B300, Nordic, 30-day lead) is exactly the profile this news favors — and sellers holding Blackwell allocation just gained negotiating room.

More from the Pulse

Related coverage

Data Center DynamicsDatacenter6h ago

Bulgarian 800 MW campus enters phase-1 commissioning

The Sofia-region AI campus, backed by a consortium of Nordic hyperscalers, will bring 60 MW online in Q3 with a further 740 MW roadmap across 2027–2029.

The RegisterAI1d ago

AMD MI355X shows 1.3× throughput vs H200 on select inference workloads

MLPerf inference results published this week show AMD's latest Instinct accelerator pulling ahead on LLM inference — though training remains an NVIDIA stronghold.