Decart Raises $300M at $4B Valuation to Challenge Nvidia’s AI Software Lock-In

Quick Facts

• Decart raised $300M at a $4B valuation, bringing total funding above $450M since founding in late 2023
• The company’s DOS software makes AI models portable across Nvidia, Google, and Amazon chips, challenging Nvidia’s CUDA lock-in
• Nvidia participated in the funding round despite Decart’s technology threatening its market dominance

Israeli AI startup Decart raised $300 million at a nearly $4 billion valuation to develop software that makes artificial intelligence models portable across different computer chips. The round was led by Radical Ventures with participation from Nvidia Corp., Adobe Ventures, Toyota Ventures and OpenAI co-founder Andrej Karpathy.

The funding brings Decart’s total capital raised above $450 million since its founding in late 2023. The company previously raised $32 million at a $500 million valuation in December 2024 and $100 million at a $3.1 billion valuation in August 2025.

Decart’s core product, the Decart Optimization Stack (DOS), allows developers to run AI models across chips from Nvidia, Google, and Amazon without being locked into a single platform. The technology directly challenges Nvidia’s CUDA software ecosystem, which has kept customers tied to its graphics processing units.

“Moving models between different hardwares is incredibly hard, and we’ve seen companies signing contracts worth hundreds of millions of dollars to help them shift models from one hardware to another,” said CEO Dean Leitersdorf. “We’re kind of the great equalizer.”

The company’s DOS 2.0 enables AI agents to process over 1,600 tokens per second, eight times the industry average. Decart claims the software reduces costs from thousands of dollars to less than $0.25 per hour for its models.

Decart also develops world models called Lucy and Oasis. Lucy modifies video streams in real-time, allowing applications like virtual try-on mirrors. Oasis generates three-dimensional environments for robot training simulations.

The company has established a strategic partnership with Amazon Web Services, becoming one of the first to deploy real-time AI models on AWS Trainium chips. Nafea Bshara, vice president of Amazon’s Annapurna Labs, said Decart’s Lucy2 model achieves over 80% Model FLOPS Utilization on Trainium3.

Decart generates revenue through licensing agreements with cloud providers and AI labs. The company claims profitability almost from launch, earning millions through its GPU optimization technology deployed by major cloud providers.

The funding comes as Israeli tech raised $4.3 billion in Q1 2026, up 18% year-over-year but across only 137 deals, the lowest deal count since 2018. The round positions Decart to compete in the infrastructure layer for next-generation AI systems.

Decart was founded by Dr. Dean Leitersdorf, 26, who earned his doctorate from Technion at 23, and Moshe Shalev. Both are veterans of Israel’s Unit 8200 intelligence unit.

Read more: Decart raises $300M for its AI optimization software, world models