NVIDIA's Rubin platform enters full production

Read the original article →

What happened

NVIDIA said its next-generation Rubin platform is in full production. Partner products built on Rubin are due in the second half of 2026. The platform includes six chips. They are the Vera CPU, Rubin GPU, NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet Switch.

NVIDIA says the platform is designed to cut training time and lower the cost of inference tokens. AWS, Google Cloud, Microsoft, and OCI are set to be among the first cloud providers to deploy Vera Rubin instances in 2026.

Why it matters

The chips that train and run AI models shape the whole industry's cost and speed. A new platform that lowers training time and token costs helps every company that builds on these chips. Early cloud support means the gains reach many users.

Demand remains huge, with frontier labs racing to secure compute.

MintedBrain take

For most users, chip news feels distant, but it sets the price and speed of the AI tools you use. Cheaper inference can mean lower prices and faster replies over time. Watch which clouds get Rubin first, since that is where new capacity will land.

References

This article was originally published at NVIDIA Blog. For the full piece, read the original article.

Discussion

  • Loading…

← Back to News