Nvidia Spends $7 Billion to Enter Open-Weight Model Race, Targeting DeepSeek and OpenAI

A report from finance.biggo.com says Nvidia has spent roughly $7 billion to enter the open-weight model race, with the headline describing the effort as taking direct aim at DeepSeek and OpenAI. The report was published on August 23.

Nvidia has long sat at the center of AI infrastructure, with its GPUs and CUDA ecosystem powering training and inference for nearly every frontier model. A multibillion-dollar push into the model layer would mark a significant expansion of that footprint.

The open-weight camp has been led in recent years by players such as DeepSeek, whose models ship with public weights that developers can download and build on, delivering near-frontier performance at relatively low cost and gaining fast traction in the global developer community.

According to the report, the move directly targets DeepSeek and OpenAI — the most prominent representative of the open route and the benchmark of closed frontier models, respectively. Entering via open weights means Nvidia is challenging both technical paths at once.

The commercial logic is straightforward: a compute vendor that also ships its own open-weight models can build ecosystem lock-in at the model layer while reinforcing demand for its own hardware.

At roughly $7 billion, the investment is on par with top-tier model research budgets, signaling a serious commitment rather than an exploratory move. For DeepSeek and OpenAI, a new entrant with chips, capital, and engineering muscle makes the frontier race significantly more complicated.

The report has not yet detailed what size or positioning Nvidia's model will have, or under what license it will be released. The key things to watch are the debut of Nvidia's open-weight model, how it integrates with the company's GPU ecosystem, and the knock-on effects on open-source communities and closed-model pricing.