Nvidia agrees to pay $6 billion for Poolside AI model software, adding a $1 billion investment and new engineering hires

Nvidia has reached a high-value licensing and partnership arrangement tied to Poolside AI model software, a deal that underscores how intensively chipmakers and AI labs are competing on model development and distribution. The company would pay $6 billion to license Poolside’s software that supports the creation and use of AI models, according to a report carried by Yahoo Finance via Quartz.

In addition to the licensing payment, Nvidia plans to invest $1 billion in Poolside. The report also says Nvidia will hire more than 100 of Poolside’s engineers to work on Nvidia’s open-weight Nemotron models.

Open-weight models are AI models whose underlying weights, rather than only a black-box API, are made available for use and adaptation. By pushing open-weight releases, a company can potentially accelerate adoption across developers and researchers who want to fine-tune the models for specific tasks, and it can also broaden the ecosystem of tools and optimizations built around the model family.

The combination of licensing software, funding the supplier, and recruiting engineers points to a strategy beyond simply purchasing model capabilities. In practice, licensing can reduce the time required to integrate proven model techniques into a vendor’s own stack. Investing and hiring can further deepen control over the model-development pipeline, help maintain compatibility with internal tooling, and retain key technical know-how that might otherwise sit outside the chipmaker’s organization.

Nemotron is part of Nvidia’s push to provide model options that align with its AI software and infrastructure. For Nvidia, the stakes are tightly linked to how enterprises and developers choose to train and run models, because workloads increasingly depend on specialized hardware and supporting software. Deals that improve access to model-building components can, in turn, influence whether customers see Nvidia platforms as a default environment for deployment.

Poolside’s role in the arrangement, as described in the report, appears to be both technical and financial. Nvidia’s $6 billion licensing payment suggests the company views Poolside’s model software as materially valuable, while the $1 billion investment indicates it is treating the partnership as longer-term rather than purely transactional. The reported plan to add more than 100 engineers further suggests an intent to embed Poolside expertise directly into Nvidia’s model engineering workstreams.

What Nvidia does not disclose in the report is as important as what it does. The available details do not specify the duration of the licensing agreement, the scope of what is covered by the software license, performance targets, geographic limitations, or whether the arrangement includes additional royalties or revenue-sharing. It also does not clarify whether Nemotron’s open-weight releases will include all components related to Poolside’s software or only those elements Nvidia chooses to make public.