Nvidia reportedly backs a $6 billion push to build an open-weight AI model with Poolside

Nvidia is reportedly entering the open-weight large language model race with a multibillion-dollar deal tied to Poolside, a move that would extend the chipmaker’s influence beyond training hardware and into model development itself. The report, published by Yahoo Finance, says the effort is sized at $6 billion and is intended to produce a “powerful open-weight AI model.” Nvidia, the report notes, is targeting systems aimed at competitors including OpenAI, Anthropic, and DeepSeek.

While Nvidia’s business has long been anchored in providing GPUs and related platforms for AI training and inference, the reported Poolside partnership indicates a different approach: supporting the creation of a model that can be more widely distributed and adapted. In the AI industry, “open-weight” generally refers to releasing the model’s parameters (the trained weights) to developers or partners, allowing more direct experimentation and fine-tuning than is typical with fully closed, black-box systems.

The potential competitive implications are straightforward. Open-weight models can lower the barrier for third parties to build applications and specialized variants, which can accelerate adoption across enterprises and developer communities. Closed-model labs typically retain the weights behind proprietary APIs. If Nvidia and Poolside deliver a strong open-weight model, it could draw developers who want flexibility, lower integration friction, or cost control through customization.

The report characterizes the project as part of a broader push by Nvidia into AI model development rather than solely selling compute. For Nvidia, this is a strategic extension of its role in the ecosystem. Even if the model is not marketed directly under Nvidia branding, a high-performance open-weight system could increase demand for the Nvidia hardware and software stacks used throughout training, tuning, and deployment.

Poolside’s name matters because partnerships often determine whether an initiative becomes a practical product or remains a research exercise. The reported deal suggests Nvidia is not attempting to build an open-weight model alone, but instead is working with an outside team that may bring model design choices, dataset strategy, and product planning. The report does not provide additional operational details in the information available here, including timelines, staffing, or release schedules.

The reported targets also frame the stakes. OpenAI, Anthropic, and DeepSeek are among the most closely watched organizations competing on performance, distribution strategy, and model safety or alignment approaches. Nvidia’s reported involvement suggests it wants to shape the terms of competition, potentially offering an open-weight alternative that pressures closed-model providers on capability and cost.

Sector context also matters for understanding why Nvidia might move in this direction. The AI market has increasingly shifted from “who has the best chips” to “who can deliver the best models and developer experiences.” Chipmakers have remained central, but control of the full stack has become a recurring theme. An open-weight model initiative could strengthen Nvidia’s position in the part of the stack that developers interact with directly.

What is still unclear, based on the material available here, is the structure of the arrangement. The report describes the deal as a $6 billion poolside-related effort and frames it as aimed at building an open-weight model. However, it does not disclose whether the model weights will be released immediately, under what licensing terms, or whether Nvidia and Poolside will have shared rights to distribution, fine-tuning, and commercial deployments. Those details are crucial for assessing how broadly the model could be used and how direct the competitive threat would be to existing closed-model offerings.

Nvidia’s next moves to watch include any official confirmations, partner details, and milestones tied to model training progress and evaluation results. If Nvidia or Poolside later publishes technical benchmarks or release plans, market participants will have more to evaluate on real performance, not just ambition.