CrossBar began by solving hard problems in semiconductor technology. As our work expanded into blockchain, digital assets, and self-custody, we realized that building secure hardware was only part of the challenge. In trustless systems, people shouldn’t have to simply trust the hardware protecting their digital assets. That realization reshaped our understanding of open source and set us on a path toward building hardware that is not only secure, but also transparent and verifiable.
1. From Secure Hardware to Trusted Hardware
For years, our focus was on advancing semiconductor technology. Through innovations in Resistive Random Access Memory (ReRAM) and secure silicon, we developed technologies that improved the performance, reliability, and security of modern hardware. Like many hardware companies, we believed that better engineering naturally led to better security.
As our work expanded into secure processors, blockchain, and digital assets, however, we encountered a different set of expectations. Traditional computing often relies on trusted operating systems, cloud providers, or device manufacturers to help establish security. Self-custody changes that model. The hardware itself becomes responsible for protecting cryptographic keys and the digital assets they secure, often without a trusted intermediary. The shift changed what users expected from secure hardware. It was no longer enough for a device to be secure by design or backed by strong engineering. People wanted more than security claims. They wanted to understand why the hardware deserved their trust and to have confidence that its protections could be independently examined rather than accepted on faith.
That realization changed the question we were asking. Instead of focusing only on “How do we build secure hardware?” we began asking, “How do users know the hardware is trustworthy?” Engineering excellence remains essential, but we came to recognize that trust also depends on transparency and independent verification.
2. Rethinking What Open Source Means
As our understanding of trust evolved, so did our view of open source. We came to see it as a practical way to make secure hardware easier to inspect and verify. Security doesn’t come from asking people to take our word for it. It comes from building systems that can stand up to independent scrutiny.
The same principle has long guided modern cryptography. It has long relied on openness and rigorous public review, demonstrating that openness strengthens security rather than weakening it. The same idea applies to the hardware that protects digital assets and sensitive data.
Rather than treating openness as an all-or-nothing decision, we take a practical approach. Licensed IP, manufacturing processes, and commercial partnerships mean that not every part of a modern chip can be made public. Instead of trying to publish everything, we open the parts of our platform that benefit most from outside review, giving researchers and developers the opportunity to examine them, evaluate them, and build on them.
For us, open-source silicon is not defined by what is published, but by what can ultimately be verified. Open source is one of the most important tools we have for building hardware that people can examine and verify for themselves.
3. From Principles to Practice
We first put these ideas into practice with Daric (the internal codename for CrossBar Chip 1), our Secure Processing Unit (SPU). Built on our ReRAM technology, Daric combines the flexibility of a modern microcontroller with the protections of a secure element. It represents our effort to build hardware that is secure by design while laying the groundwork for greater transparency.
Our collaboration with Andrew “bunnie” Huang, a pioneer in open hardware and hardware security research, further advanced that vision. Alongside contributing an open-source RISC-V core, Bunnie played a key role in making the chip’s RTL publicly available wherever third-party IP constraints allowed. We also open-sourced the hardware and software drivers for Daric on GitHub, extending that commitment to openness beyond the silicon itself.
Publishing the design, however, addresses only part of the trust problem. It allows people to examine how a chip was intended to work, but not whether the manufactured silicon faithfully reflects that design. To help bridge that gap, our collaboration also led to BaoChip-1x, a mostly open variant that shares the same 22 nm production silicon as Daric. Built on the same physical chip rather than requiring a separate tapeout, BaoChip-1x provides an open platform for research and software development while allowing the broader community to evaluate the same hardware foundation used in commercial products.
That shared hardware foundation also made BaoChip-1x an ideal platform for advancing IRIS (Infra-Red, in situ) inspection. Developed by Bunnie, IRIS is a practical, non-destructive technique for inspecting fabricated silicon, helping increase confidence that manufactured chips faithfully reflect their intended design. Together, BaoChip-1x and IRIS help narrow the gap between open designs and trustworthy hardware by making silicon-level verification more practical.
Daric, BaoChip-1x, and IRIS represent different stages of the same vision. Daric provides a secure hardware foundation, BaoChip-1x opens that foundation to researchers and developers, and IRIS brings practical verification to the manufactured silicon. Together, they move us closer to hardware that is not only secure but also transparent and verifiable.
4. What’s Next for Trusted Hardware
The future of trusted hardware will not be shaped by any one company or technology alone. Open hardware continues to evolve, and important challenges remain. Independent verification is still difficult, development tools continue to mature, and balancing openness with the realities of commercial semiconductor design requires ongoing innovation.
Meaningful progress will require researchers, hardware companies, and open-source developers to share knowledge, review each other’s work, and keep improving the tools the ecosystem depends on. Open hardware is still evolving, and so are we. We’ll keep building secure hardware, publishing more of our work when appropriate, and making it easier for others to inspect and test what we build. In the end, trust is strongest when it can be verified.
— The CrossBar Team