Nvidia Expands Nemotron Open-Weight Push to Counter Chinese Labs Under B Poolside Deal

Nvidia Expands Nemotron Open-Weight Push to Counter Chinese Labs Under $6B Poolside Deal Nvidia plans to use the infrastructure and engineering team acquired through its $6 billion deal with AI startup Poolside to build frontier open-weight models under its Nemotron family, according to reporting from the Wall Street Journal. The initiative aims to counter the rapid global adoption of Chinese open-weight systems like DeepSeek-V3, Moonshot AI's Kimi K3, and Alibaba's Qwen series, while offering

2 min
Nvidia Expands Nemotron Open-Weight Push to Counter Chinese Labs Under B Poolside Deal

Nvidia Expands Nemotron Open-Weight Push to Counter Chinese Labs Under $6B Poolside Deal

Nvidia plans to use the infrastructure and engineering team acquired through its $6 billion deal with AI startup Poolside to build frontier open-weight models under its Nemotron family, according to reporting from the Wall Street Journal. The initiative aims to counter the rapid global adoption of Chinese open-weight systems like DeepSeek-V3, Moonshot AI's Kimi K3, and Alibaba's Qwen series, while offering enterprise customers a customizable alternative to closed APIs from OpenAI and Anthropic.

Under the transaction terms first disclosed in an investor letter, Nvidia is paying $6 billion to license Poolside's "Model Factory" training and evaluation software and is extending employment offers to 109 staff members who developed Poolside's Laguna coding models. In addition, Nvidia is investing $1 billion directly into Poolside at a $12 billion pre-money valuation.

Strategic Pivot to Open-Weight Competition

The transaction reflects an intensifying battle between proprietary closed-source providers and open-weight model developers. Over the past twelve months, Chinese labs have narrowed the capability gap with American frontier models across code generation, reasoning, and multimodal benchmarks, distributing weights freely under permissive licenses.

By bringing Poolside's Model Factory tooling and specialized pre-training team into Nvidia, the chipmaker intends to scale its Nemotron series into full-scale frontier competitors. While Nvidia's primary business remains accelerated compute silicon and systems, releasing competitive open weights ensures enterprise developers remain anchored to CUDA software pipelines and Nvidia hardware configurations rather than migrating to alternative cloud infrastructures.

Nvidia Model Factory Architecture

Structure of the Non-Exclusive Licensing Agreement

The arrangement between Nvidia and Poolside mirrors previous non-exclusive transactions Nvidia has completed to secure critical AI software and talent without triggering formal antitrust reviews:

  • Model Factory Licensing: Nvidia gains non-exclusive access to Poolside's proprietary model-building pipelines, synthetic data engines, and RL verification environments.
  • Team Integration: 109 engineers and researchers behind the Laguna model series are transitioning to Nvidia's foundation model teams.
  • Corporate Independence: Poolside's three co-founders remain with the startup, which will continue independent operations.
  • Capital Distribution: Poolside plans to distribute the $6 billion licensing payment to its venture investors by the end of 2027.

In its letter to investors, Poolside noted that continuing to compete independently in frontier foundation model pre-training would require access to computational scale that exceeded the startup's standalone resources.

Hardware Ecosystem and Enterprise Customization

Enterprise demand for open-weight models has grown due to data privacy constraints, lower inference costs, and on-premises deployment requirements. Open-weight checkpoints allow organizations to fine-tune weights on proprietary data, deploy within sovereign cloud boundaries, and implement custom speculative decoding or quantization schemes.

Nvidia's accelerated push into open weights positions the company as both the dominant supplier of AI hardware and a primary provider of the foundational software layer running on top of it.

Sources

Written by

More to read

  • AI Workflow Startup Relay Shuts Down as Team Joins Google Chrome to Build Browser Agents

    AI-driven workflow automation startup Relay is shutting down its independent product operations, with founder and chief executive officer Jacob Bank and key engineering staff joining Google's Chrome division to develop browser-native AI agent capabilities. Relay, founded in July 2021 to compete with legacy workflow platforms like Zapier through generative AI integrations, raised $8.1 million across two venture funding rounds. The company phased out free tier access on August 15, 2026, and will

    1 min
  • Groq Secures 50M at .5B Valuation to Expand Nvidia-Powered AI Neocloud

    AI infrastructure provider Groq has raised $350 million in a Series A funding round at a $3.5 billion valuation, led by investment firm Disruptive with expected participation from Nvidia subject to customary closing conditions. The financing accelerates the company's structural pivot from developing custom inference silicon toward operating an enterprise-grade inference cloud powered by Nvidia accelerated computing systems. The round follows a $650 million capital raise completed in June 2026 a

    1 min
  • Infinite Agentic Loops in Production: Architecture, Feedback Topologies, and Bound Verification

    Autonomous AI agents have transitioned software architectures from static, single-turn request-response patterns into stateful, iterative execution loops. Built around foundational paradigms such as ReAct (Yao et al., 2022) and implemented across frameworks including LangGraph, CrewAI, AutoGen, and the OpenAI Agents SDK, agents repeatedly perceive environmental state, reason over intermediate goals, dispatch tool invocations, observe execution outputs, and append new observations back into their

    1 min