SUMM

Diode recently closed an $11.4 million Series A led by A16Z, just a year after their YC Summer 24 batch
Diode uses AI to automate circuit board (PCB) design, functioning as an AI-enabled design shop for printed circuit boards
Their customers include Fortune 100s, robotic foundation model startups, autonomous boat companies, and smaller startups

Founders met during COVID; one transitioned from software to hardware engineering during college
Both had exposure to hardware and software product development, observing that hardware lacked software-like tooling to quickly detect and fix mistakes
Discovered hardware development is slower and more error-prone compared to software due to lack of effective tooling

Original YC application centered on AI-powered verification of circuit board designs to catch errors, similar to a software compiler
Built generative pipelines for creating boards and injecting mistakes to train AI, but customer feedback showed minimal interest in such verification tools
After 100+ customer conversations, realized the bigger market pain point was in generating new designs, not in verifying existing ones

Through YC, networked with batchmates and discovered significant demand for the automatic generation of custom development boards (e.g., for robotics companies using Nvidia Jetsons)
Quickly landed their first customer after this discovery, validating the need for full end-to-end PCB design solutions rather than component tools
Identified that the shortage of skilled PCB designers made this offering particularly valuable for both startups and established companies

Initial verification pipeline work led them to develop robust tools for generating boards, which became a core offering
Transitioned from verification services to delivering complete, verified board designs, using AI internally to automate and accelerate the process
Take full ownership of delivered work, verifying outputs before customer handover, similar to traditional product development but faster

Reframed PCB design as a software problem, adapting tools and verification practices from software engineering to hardware design
Used AI to automate design while maintaining quality standards; achieved significant productivity gains by applying software best practices to hardware

Modern AI models possess substantial knowledge about electrical engineering but haven’t been used in PCB design due to traditional, visual-centric tools
Diode intermediates by converting designs to code, allowing models to design efficiently, then exporting to visual representations for human users
Enables domain experts to review outputs in familiar formats without needing to learn new coding paradigms

Two-person team designed over 100 boards in a few months, vastly outpacing traditional electrical engineering output
Internal division: one founder leads electrical engineering, the other leads software, with custom software tools enabling speed
Emphasizes building tools on top of open-source electronics software for further efficiency gains

Developed a custom language for schematics that is both LLM-friendly and human-readable
Infrastructure designed for modular deployment and strict security requirements (e.g., air-gapped aerospace systems)
Core logic implemented in Rust with browser-based visualizations and real-time schematic compilation
End-users can configure board modules interactively and see instant updates

Aim to 10x hardware design speed by applying AI advances from software to physical world challenges
Hardware design seen as the next major challenge/opportunity, attracting engineers looking for impactful, difficult problems
Diode is hiring: seeking curious, experimentation-oriented software engineers comfortable with research problems and building customer-facing interfaces
Prioritize shipping practical products while advancing the technical frontier

How Diode Is 10x-ing Hardware Design