Sakhanda Wire
NVDA $154.51 -1.04% MSFT $506.49 -2.22% GOOGL $281.27 -0.86% META $621.51 -0.78% AMZN $256.90 -0.33%
← Back to the news

NVIDIA automates supply allocation with Palantir and cuOpt

NVIDIA is using Palantir Foundry and the GPU-accelerated optimisation library cuOpt to automate hardware allocation decisions across manufacturing sites worldwide. The company tracks delivery from wafer-out to the first token, separating the measure into time-to-rack and time-to-token. The latter includes power, cooling, networking and software readiness on the first day.

A Grace Blackwell NVL72 rack contains 18 compute trays. Each tray needs two Grace CPUs, four Blackwell GPUs and 32 HBM3e memory packages sourced through thousands of suppliers and partners. The supply network planned for Vera Rubin is twice the size of the Grace Blackwell network. NVIDIA revises factory allocations weekly across rolling two-quarter periods, accounting for component availability, production limits and customer schedules.

Foundry’s Ontology connects facilities, supplier commitments, inventories and production targets. cuOpt uses this information in a mixed-integer linear program intended to reduce “Time of Ownership”, the period between material receipt and shipment of completed sub-assemblies. NVIDIA also post-trained Nemotron 3.5 Lightning on anonymised operational records. On historical allocation data, it reached 86.7 percent decision accuracy, compared with 55.5 percent for Nemotron 3 Ultra and 17.5 percent for the untuned Lightning model. NVIDIA plans to use recorded decisions and outcomes for future reinforcement-learning preference data.

Written by Sakhanda Wire AI from reporting by AI News. Facts, quotes and full context are in the original article below.

Originally published by AI News on

Read the original on AI News ↗

The reporting and the image belong to AI News and are credited and linked above; the text on this page is Sakhanda Wire's own summary.

← Back to the news

More stories

All the latest news