Nvidia, Dell Back $200M Series A for AI Optimizer doubleAI

Tel Aviv-based artificial intelligence developer doubleAI announced today that it has secured $200 million in Series A funding. The investment round was co-led by Lightspeed Venture Partners and Bessemer Venture Partners, with participation from Nvidia Corp., Dell Technologies Inc., and several other undisclosed investors.

According to reporting by CTech, this latest capital injection values the company, officially known as AA-I Technologies Ltd., at more than $1 billion. The funding supports the development of Maxwell, an AI model designed to automatically improve the efficiency of software code. The firm plans to sell this technology to enterprises seeking to reduce the hardware requirements for their AI workloads.

Technical Capabilities and Performance

Maxwell distinguishes itself by optimizing software binaries rather than modifying source code. This approach allows the system to apply improvements to applications while they are running in production environments, removing the need for developers to manually rewrite program logic. The model analyzes how a program interacts with processed data and underlying chips to identify optimization opportunities.

The system supports both graphics processing units and central processing units. In March, doubleAI demonstrated Maxwell’s capabilities by rewriting Nvidia’s open-source cuGraph library, which enables customers to use graphics cards for graph analytics. The company stated that the model sped up 55 percent of the algorithms in the library by a factor of two or more.

Maxwell also created an optimized version of pandas, a widely used open-source data science toolkit. According to the company, this optimization improved the library’s performance on the PDS-H query speed benchmark by a factor of 38. More recently, doubleAI used the model to double the video generation speed of a world model called LTX-2.5.

Broader Applications and Recent Milestones

Beyond software optimization, doubleAI has highlighted other recent achievements involving its technology. Earlier this month, the Maxwell model broke a long-standing chess challenge. Around the same time, an agent developed by doubleAI made a new mathematical discovery, demonstrating the model’s versatility across different complex tasks.

The company’s technology addresses a fundamental aspect of computing: processors do not run source code directly but instead execute binaries, which are machine-generated translations of programs into simple computing operations. By targeting these binaries, doubleAI aims to provide a seamless way to enhance application efficiency without disrupting existing development workflows or requiring extensive manual intervention from engineering teams.

What should you do?

Security and technology professionals should monitor developments in AI-driven code optimization tools like doubleAI’s Maxwell. As these models gain traction, organizations must evaluate how binary-level modifications might impact their software supply chain integrity and change management processes.

What to do and how to stay safe: doubleAI

• Review your software development lifecycle policies to determine how binary-level optimizations are validated and tested before deployment.
• Assess whether your current monitoring systems can detect performance changes or behavioral shifts resulting from AI-optimized binaries.
• Evaluate the security implications of allowing third-party AI models to modify production software without source code access.
• Stay informed about updates from vendors like Nvidia and Dell regarding their partnerships with AI optimization firms.

Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation

General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

Who invested in doubleAI’s latest funding round?

Lightspeed and Bessemer co-led the $200 million Series A round, with participation from Nvidia, Dell Technologies, and other investors.

How does doubleAI’s Maxwell model optimize software?

Maxwell optimizes software by modifying binaries rather than source code, allowing it to improve efficiency without requiring manual developer changes.

What performance improvements has Maxwell demonstrated?

Maxwell sped up 55% of algorithms in Nvidia’s cuGraph library by two or more times and improved pandas performance on PDS-H benchmarks by a factor of 38.