JuliaHub Raises $65M Series B, Launches Dyad 3.0 for Industrial AI

The industrial engineering sector is finally catching up to the software revolution, and JuliaHub is leading the charge with a $65 million Series B funding round announced April 30, 2026. The capital injection, led by Dorilton Capital with participation from General Catalyst, AE Ventures, and former Snowflake CEO Bob Muglia, coincides with the launch of Dyad 3.0, the company's updated AI platform for designing complex physical systems.

According to the official press release, Dyad 3.0 represents what the company calls "agentic AI" for hardware engineering. The platform is designed to compress R&D cycles from months to days or even minutes for systems ranging from heat pumps to satellites to semiconductors. Several Fortune 100 companies are already using Dyad across aerospace, government, automotive, HVAC, and utilities sectors.

Here's the thing: hardware engineering has been stubbornly resistant to AI adoption. While developers have been using tools like Claude Code and Gemini for years, industrial engineers still wrestled with legacy systems that required manual physics modeling, control logic development, and safety analysis. Dyad attempts to bridge this gap by combining physics-based simulations, control systems, safety analysis, and code generation in one environment.

The platform creates "digital twins"—virtual models of real-world systems that can be tested and improved using AI. It uses scientific machine learning (SciML) to continuously refine these models based on real-world data, helping engineers predict failures, improve efficiency, and optimize performance. The key differentiator: Dyad ensures AI-generated designs follow the laws of physics, which matters when an error means a bridge collapse or battery fire rather than a software bug.

Viral Shah, CEO of JuliaHub, framed the vision bluntly: "It's not about helping engineers complete one small task at a time. It's agentic engineering at scale, where teams can feed a full specification to Dyad and have it design the complete system. Spec in. Design out." The platform automates complex engineering tasks like building control systems for chemical plants, work that traditionally takes weeks.

The funding announcement comes at a critical inflection point. McKinsey estimates that a cumulative $106 trillion in investment will be necessary through 2040 to meet infrastructure needs. Engineers planning these updates need tools that move at the pace of AI-enhanced software. Dyad 3.0 builds on Dyad 1.0 (launched June 2025) and Dyad 2.0 (December 2025), connecting autonomous agents with scalable physics simulations and the ability to generate code for embedded systems.

Daniel Freeman, who led the Series B round for Dorilton Capital, noted that systems modeling is one of the most strategically important layers of the AI-native engineering stack. "JuliaHub has built something extraordinary with Dyad: a platform that doesn't just model systems, but compiles them, taking engineers from concept to production control code in a single environment," Freeman said in the release.

The platform has already attracted partnerships with major players. Prith Banerjee, Senior Vice President of Innovation at Synopsys, commented on the integration with Ansys TwinAI, noting that Dyad enables high-fidelity hybrid digital twins by integrating physics-based simulation with data-driven models. What once required extensive manual effort can now be done far more efficiently.

Recent agentic benchmarking for chemical process modeling showed general LLM systems like Codex, Claude Code, and Gemini barely completed initial setup. Dyad almost entirely automated creating model-predictive controllers to optimize yields of a chemical plant. The distinction matters: general-purpose AI cannot guarantee a model obeys physics, but Dyad's modeling language is purpose-built for this constraint.

The physical reality of using Dyad involves engineers feeding specifications into a cloud-based workspace where autonomous agents scan scientific knowledge to improve models. AI-automated lab testing ensures models match physical reality. Streaming data mixed with SciML allows models to automatically grow as the system learns from the real world. Engineers don't have to write every line of code to try millions of designs while maintaining oversight to ensure safety.

David Joyce, former CEO of GE Aviation and Vice Chair of GE, called it a disruptive transition in engineering system design software. "With Dyad, you can model the physics, develop control algorithms with auto code generation, and create accurate digital twins and surrogates for rapid development of deep learning inference models, all enabled by AI." (Whether this translates to actual productivity gains in real-world deployments remains to be seen.)

The $65 million raises questions about market timing and adoption barriers. While the technology promises dramatic efficiency gains, industrial sectors have notoriously long procurement cycles and rigorous safety requirements. Whether Fortune 100 companies will actually pay premium prices for AI-generated designs that must pass regulatory approval is the real test.

JuliaHub's bet is that the infrastructure investment wave through 2040 creates enough demand to overcome traditional engineering conservatism. The company has positioned itself at the convergence of physics, control logic, and AI—a space where errors have real-world consequences but the productivity gains could be transformative.

Whether users actually pay for it remains the real question.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn