Davide Asnaghi

Davide Asnaghi runs Diode Computers, a company built on a stubborn observation: the chips inside a modern device can be verified transistor by transistor, billions at a time, yet the software that lays out the copper connecting them still lets engineers ship simple, expensive mistakes. Diode's answer is to treat a circuit board less like a drawing and more like a program - something you write, compile, and check in milliseconds.

The pitch is deceptively plain. Instead of dragging shapes around in decades-old EDA tools, a Diode engineer describes a board in Python. A compiler written in Rust turns that description into a real, manufacturable layout, applies the constraints of a specific factory, and exports to the formats the rest of the industry already uses - Altium, Cadence, ODB++, IPC-2581. Verification that used to arrive days later, after a board came back wrong, now happens while you type.

That gap is what Diode is selling against. The company is backed by Y Combinator and, in 2025, closed an $11.4 million Series A led by Andreessen Horowitz. Early customers include Physical Intelligence and Saronic, names that signal where Asnaghi is aiming: robotics and defense-adjacent hardware teams that need to iterate fast and can't wait weeks for a board revision. By the company's own account, a single engineer on the platform shipped more than 100 unique designs in a year.

Teaching an AI to be an engineer

The most ambitious part of the plan reaches past tooling. Diode is working with Anthropic to make Claude a more capable electrical engineer - not a chatbot that talks about circuits, but a model that can help produce boards that actually ship. The logic follows from the architecture: if a design is code, and a compiler can instantly tell you whether it's valid and manufacturable, then an AI has a tight feedback loop to learn inside. Diode also trains smaller models with reinforcement learning to catch the kinds of design errors that slip past human review.

Asnaghi frames the destination bluntly.

From Italy to the factory floor

Asnaghi grew up in northern Italy and spent a year in Minnesota as a teenager, an early collision of cultures that shows up later in how he thinks about manufacturing. He studied engineering, coming to UC Berkeley - by way of Politecnico di Milano - and finishing a master's in 2019 focused on prosthetic devices and brain-machine interfaces. Before circuit boards, his hardware was human-scale: he co-founded Project Sparthan, an open-source effort to make affordable 3D-printed prosthetic hands for children, a collaboration between Berkeley and the Hong Kong University of Science and Technology.

After Berkeley he joined Apple's Special Projects Group, working on robotics, then built hardware at medical-device companies including Butterfly Network and Chromatic. Along the way he spent real time in Hong Kong and Shenzhen, not as a tourist but studying how Asia's electronics manufacturing ecosystem actually runs. That experience produced one of his more pointed arguments - that the usual story about why China dominates hardware is out of date.

The takeaway he draws from that is optimistic rather than fatalistic: if automation, not cheap labor, is the real edge, then better software and better tooling can bring competitive circuit board work back to the United States. Diode's open-source design tools are the opening move - lower the cost and friction of designing a board, and the case for making it domestically improves.

The copper merchant

Asnaghi jokes that he's a "copper merchant," a nod to the humble metal at the center of every board he cares about. The self-deprecation hides a consistent thread across his career. Prosthetic hands, robots, medical devices, and now circuit boards all share the same shape of problem: complex physical hardware that too few people can build, held back by tools that haven't kept pace. His response each time has been to make the hard thing more accessible, and to open-source as much of it as possible.

Diode is the largest expression of that instinct so far. If the bet pays off, hardware development starts to feel less like a series of slow, risky commitments and more like the write-compile-run rhythm software engineers take for granted. If it doesn't, Asnaghi will still have pushed on one of the most overlooked bottlenecks in modern manufacturing - the unglamorous layer of software that decides how quickly a good idea becomes a board you can hold.