doubleAI Raises $200M for Binary Optimisation Model Maxwell

Tel Aviv-based doubleAI has raised $200 million in a Series A funding round, with participation from Nvidia and Dell Technologies. According to SiliconANGLE, Lightspeed and Bessemer co-led the investment, which values the company at more than $1 billion. The firm, officially known as AA-I Technologies Ltd., plans to sell its AI model to enterprises seeking to reduce hardware usage for AI workloads.

How it works

doubleAI developed an AI model called Maxwell that automatically improves software efficiency. Unlike traditional methods that require developers to modify source code, Maxwell applies optimisations directly to the binary. This machine-generated translation represents a program as simple computing operations. The approach allows for modifications while applications run in production, removing the need for manual source code adjustments by engineering teams.

Performance examples

Maxwell analyses how a program interacts with processed data and underlying chips to find optimisation opportunities. The model supports both graphics processing units and central processing units. In March, doubleAI revealed that Maxwell rewrote Nvidia’s cuGraph library to lower hardware usage. The model sped up 55% of the algorithms in that library by a factor of two or more, demonstrating its capability to enhance graph analytics performance.

Software applications

The technology has been applied to several open-source projects and benchmarks. Maxwell created an optimised version of the pandas data science toolkit, improving performance on the PDS-H query speed benchmark by a factor of 38. Additionally, doubleAI used the model to double the video generation speed of a world model called LTX-2.5. These examples illustrate the range of software components that can be processed by the system.

Vendor strategy

• Sell the model to enterprises interested in lowering AI workload hardware usage.
• Support both graphics processing units and central processing units.
• Apply optimisations to binaries rather than source code.
• Enable modification of applications while they are running in production.

Why it matters

Binary optimisation changes how software runs without altering the original code. Teams usually rely on source code for security audits, but modified binaries can bypass these checks. It is essential to understand that performance gains may come with hidden risks. Check whether your monitoring tools can detect changes in compiled files. Ensure that any automated adjustments are logged and reversible. This shift requires new validation methods for software integrity.

What to do and how to stay safe: doubleAI

• Audit your software supply chain to identify where binary modifications occur.
• Implement integrity checks for compiled applications before and after deployment.
• Review vendor documentation for tools that modify binaries in production.
• Ensure logging captures all changes to executable files in your environment.

Step-by-step guide: Stop Sensitive Data Leaks in AI Chatbots: Priority Controls

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

Frequently asked questions

Who invested in doubleAI?

Lightspeed and Bessemer co-led the round, joined by Nvidia, Dell Technologies and others.

What does the Maxwell model do?

It optimises software binaries for GPUs and CPUs without requiring source code changes.

How much funding did doubleAI raise?

The company raised $200 million in its Series A round.