Urgent.News

What's breaking now, across thousands of outlets.

AI

Token sticker shock pushes enterprises from frontier labs toward open-weight AI factories

Enterprises have spent the AI boom racing to secure GPUs, but the harder decisions now sit above the hardware. Token bills, data sovereignty and model choice increasingly decide which AI pilots reach production, a shift on display at the SC26 supercomputing conference. Dell Technologies Inc. has spent the past year pushing its AI Factory beyond […] The post Token sticker shock pushes enterprises…

Token sticker shock pushes enterprises from frontier labs toward open-weight AI factories

Enterprises are shifting their focus from securing GPUs to considering token costs, data sovereignty, and model choice as AI pilots move toward production. This shift is evident at the SC26 supercomputing conference, where Dell Technologies has been promoting its AI Factory solution. Brad Maltz, senior director of AI solutions at Dell, emphasizes that hardware is crucial but that a full-stack AI factory approach, encompassing software, open-source tools, models, and operations, is needed.

Token bills and data sovereignty are driving enterprises towards open-weight models and a full-stack AI Factory, rather than relying solely on expensive hardware. Despite pilot mode still being common, enterprises are beginning to consider the significant token spend, often exceeding $10,000 per person per month. Data sovereignty, intellectual property leakage, security, and governance are additional factors that must be addressed before a pilot goes live.

This is prompting many organizations to explore open-weight models that can be fine-tuned and run on-premises, such as Nvidia's Nemotron in the U.S. Dell has launched its AI Ecosystem Program to validate software partners on its AI Factory, with about 60 more partners applying or under evaluation. This expansion will enable Dell to address 10 to 30 more customer problems.

Ultimately, Dell views data as the lifeblood of AI, emphasizing the importance of keeping models and data under enterprise control, often through disconnected, on-premises systems.

Written by urgent.news from SiliconANGLE's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at siliconangle.com →