#USA #AI #Cambridge #JuliaHub #Dyad 3.0

JuliaHub Secures $65 Million in Series B Funding and Launches Dyad 3.0

Cambridge, Massachusetts - JuliaHub, a prominent name in the field of scientific AI, has announced an exciting development in their portfolio with the launch of Dyad 3.0. This news comes alongside their successful completion of a $65 million Series B funding round, primarily led by Dorilton Capital, with significant contributions from General Catalyst, AE Ventures, and technology investor Bob Muglia, a former CEO of Snowflake.

Revolutionizing Industrial Design

Dyad 3.0 is positioned as the world's first autonomous AI platform for hardware engineering, designed to accelerate the development and testing processes of complex systems. This innovative platform enables engineering teams to reduce research and development times from months to mere days, streamlining the design cycles for everything from heat pumps to space satellites and semiconductors.

Several Fortune 100 companies across various sectors, including aerospace, government, automotive, HVAC, and utilities, are already harnessing Dyad's capabilities to enhance their operational efficiency. Daniel Freeman from Dorilton Capital emphasized the strategic importance of system modeling in AI-native engineering architecture, noting how JuliaHub's Dyad is fundamentally changing how physical systems are designed and constructed.

Addressing the Challenges of Hardware Innovation

JuliaHub’s Dyad tackles what is recognized as a significant barrier to innovation in the hardware sector, which has not yet fully embraced the AI revolution seen in software development. Despite the significant advances in tools like Codex and Gemini for software, industrial engineers continue to grapple with outdated tools.

According to McKinsey, a staggering investment of $106 trillion will be necessary by 2040 to meet the demand for new and updated infrastructure. Engineering teams responsible for planning and constructing these upgrades will require solutions that enable them to match the pace of AI-enhanced software development. Here is where Dyad shines, providing teams with an AI-based environment for modeling, testing, and validating industrial systems - a perfect analogy to how software like Claude Code is applied to the physical world.

Dyad 3.0 builds upon the releases of Dyad 1.0 and Dyad 2.0, which debuted in mid-2025 and late 2025, respectively. It effectively connects autonomous agents to scalable physical simulations, rigorous controls, safety analyses, and the ability to generate code for embedded systems. As a result, generating highly-detailed digital twins and refining controllers for specialized deployment scenarios can now be achieved without requiring a specialized academic background.

Scaling Engineering Automation

JuliaHub’s CEO, Viral Shah, emphasizes that their aim with Dyad is not simply to assist engineers with isolated tasks but rather to facilitate automated large-scale engineering. Dyad enables teams to provide complete specifications, which the platform then utilizes to design full systems, demonstrating a remarkable leap in engineering automation.

The platform’s cloud-based AI agents continuously scan global scientific knowledge, enhancing the accuracy and relevance of models. AI-assisted automated laboratory tests are expanding in scope to ensure that models align closely with real-world physical conditions. This synergy between data transmission and Scientific Machine Learning (SciML) allows models to evolve autonomously in response to real-world learning. The robust ecosystem and simulation language of Dyad provide a foundation for ensuring process verification and that assumptions are aligned with client requirements.

A Pioneering Step Forward

Prith Banerjee, Senior Vice President of Innovation at Synopsys, remarked on the transformational potential of Dyad, noting how it seamlessly integrates scientific AI, automated modeling, and robust build pipelines. Coinciding with this integration, Dyad provides hybrid digital twins by merging physics-based simulations with data-driven models, thereby significantly amplifying efficiency across the engineering lifecycle and redefining the design and validation processes for software-defined intelligent systems.

One of the most critical aspects emphasized by the JuliaHub team is the recognition that general-purpose AI cannot ensure adherence to physical laws in modeling. In physical engineering, a mistake is not simply a minor setback; it could lead to catastrophic failures, such as collapsed bridges or battery fires. This understanding underpins Dyad's development because it ensures that the AI's framework is inherently grounded in the laws of physics.

In practical applications, in collaboration with Binnies and Williams Grand Prix Technologies, JuliaHub has developed a SciML-based digital twin capable of predicting pump failures in water distribution systems with an impressive 90% accuracy, utilizing just four sensor inputs. This ability to switch operations from reactive measures to predictive decision-making illustrates Dyad's extraordinary value in real-world applications.

Join Us for the Live Launch Event

JuliaHub is set to officially present Dyad 3.0 at a live event on May 19, demonstrating product capabilities and showcasing testimonials from clients spanning various industries, such as aerospace, HVAC, utilities, and robotics. Attendees will witness firsthand how Dyad enhances operational efficiency and safety.

For more information and media inquiries, please contact [email protected].

About JuliaHub

Founded in 2015 by the creators of the high-performance open-source programming language Julia, developed at MIT, JuliaHub has emerged as a leader in scientific AI. The company aims to equip those tackling some of the world's most complex scientific and technical challenges with cutting-edge AI tools in a secure and seamless environment. JuliaHub combines advanced mathematical computing and machine learning expertise to enable SciML techniques, digital twin solutions, and state-of-the-art modeling and simulation across various industrial sectors.