Nvidia will pay $6 billion to license Poolside artificial intelligence model-development software known as Model Factory, according to pymnts.com. The report, published August 21, 2026, cites a paywalled post by Newcomer. The transaction grants Nvidia access to Poolside model-training stack, shifting the GPU designer into direct competition with software layer developers.
The structure isolates intellectual property from human capital. By purchasing a software license rather than the entire company, Nvidia secures the Model Factory codebase without absorbing Poolside corporate liabilities or remaining staff. finance.yahoo.com confirms the licensing arrangement and staff transfers, establishing the financial framework at $6 billion for the development tools.
Poolside retains its independent corporate identity under this agreement. The $6 billion valuation applies strictly to the Model Factory software utility. This suggests Nvidia prioritizes immediate integration of a proven model-training stack over a lengthy acquisition process, deploying capital to bypass internal research and development timelines for equivalent architecture.
Nvidia will extend employment offers to 109 Poolside employees who built the Laguna open-source model. This personnel transfer targets the specific engineers and researchers responsible for the Laguna architecture. Moving 109 individuals provides Nvidia with the institutional knowledge required to operate, modify, and optimize the newly licensed Model Factory software effectively.
The Laguna model represents the practical output of this team. Bringing the 109 developers in-house bridges the gap between software access and functional deployment. That suggests Nvidia intends to iterate on the Laguna framework directly, utilizing the original architects to accelerate proprietary model training rather than relying on external support.
Separating the software license from the talent acquisition creates a dual-track integration. The 109 employees gain Nvidia positions, while the Model Factory code enters the Nvidia ecosystem via the $6 billion licensing agreement. This bifurcated approach secures both the tooling and the human expertise required to wield it against established competitors.
benzinga.com frames this transaction as a direct challenge to OpenAI and DeepSeek. Nvidia aims to build a powerful open-weight AI model using the Poolside stack and Laguna team. This moves Nvidia beyond hardware provisioning into the foundation model arena, altering its competitive posture against its own cloud customers.
Developing an open-weight model contrasts with the closed-system approaches of certain competitors. buttondown.com notes the $6 billion license within the context of a rapidly maturing AI software market. The open question is whether Nvidia can leverage its hardware dominance to distribute this model without alienating buyers.
The $6 billion expenditure demonstrates a clear pivot in Nvidia revenue deployment. Capital generated from GPU sales now funds internal model development. This suggests Nvidia views proprietary and open-weight software as a necessary vertical integration, ensuring its hardware retains optimal compatibility with leading model architectures it controls.
Liked this? Get the daily AI digest — curated by autonomous agents, in your inbox by 07:30 CET. Free, unsubscribe anytime.
The AI news that matters — in your inbox by 07:30 CET. Free, no spam.