Booz Allen Switches to NVIDIA's Nemotron Large Model

On May 29, 2026, at the Ronald Reagan Presidential Library in Simi Valley, California, the 2026 Reagan National Economic Forum was held, attended by Horacio Rozanski, CEO of Booz Allen Hamilton.

NVIDIA is facing the most severe reputational blow in its 30-year history, triggered by reports of an extramarital affair involving CEO Jensen Huang. However, the AI chip giant is also seeing a silver lining: its open-source large model project has shown initial progress.

Booz Allen Hamilton is a government contracting service provider with a market capitalization of $10 billion. The company's Chief Technology Officer, Bill Vass, revealed that months ago, the underlying model used for cybersecurity and defense business software switched from primarily using the Anthropic model to the NVIDIA Nemotron model.

Vass stated that this model replacement significantly reduced costs, but the company declined to disclose the specific amount saved. According to insiders, Booz Allen invests tens of millions of dollars annually in AI projects.

In an interview in August, Vass mentioned, "Our current approach is exactly the opposite of 'token maximization.'" "Token maximization" refers to the practice of consuming as many AI token resources as possible.

Vass noted that during the software development phase (as opposed to the software runtime phase), most of Booz Allen's developers are still using Anthropic's Claude-Code, unless clients specify other coding tools. This is likely because Claude-Code is still recognized as the best-performing tool in the software engineering field. However, Vass indicated that as Nemotron's capabilities continue to improve, the company intends to encourage internal developers to try using it for coding.

Thanks to its collaboration with Palantir, Nemotron is gradually opening up the market among government contractors and agencies. Recently, the partnership between the two companies deepened: both positioned themselves as AI providers to help enterprises reduce reliance on cutting-edge large model laboratories like Anthropic and OpenAI. NVIDIA has also recently signed a large agreement to use Palantir's software internally to manage its supply chain.

At the end of June, Palantir released a product that facilitates U.S. government agencies and enterprises in accessing NVIDIA's open-source models through the Palantir platform. Palantir's CEO revealed this summer that some government clients have switched from closed-source cutting-edge large models to Nemotron.

Not only large clients are trialing Nemotron. Data from the open-source code platform Kilo-Code shows that in October, the platform processed approximately 50 trillion tokens, with the NVIDIA Nemotron model accounting for nearly 17%, up from 8% the previous month. Individual developers and developers from small and large enterprises mainly use Kilo-Code to access various open-source models for programming tasks.

Despite these advancements, NVIDIA's model overall still lags behind mainstream cutting-edge large models. In the past month, Nemotron accounted for only about 4% of the total token processing volume on the OpenRouter platform, a slight decline from a peak of 7% in a week in August. OpenRouter, which belongs to Stripe, allows businesses to switch between different large models with one click; it processes about 700 trillion tokens monthly, with task types broader than just programming.

So far in November, on the OpenRouter platform, Anthropic accounted for 31% of total customer spending, ranking first among all AI service providers, while NVIDIA accounted for less than 1%.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.