This is a classic case of mistaking open source for commodity and underestimating what matters most in semis integration, IP licensing, qualification, and commercial execution.
Yes RISC-V is open source. So is Mbed. And yes, there are compression algorithms and off the shelf blocks anyone can technically access. But that doesn’t mean you can just stitch them together and get a 16nm edge AI SoC with ultra-low power performance, deep sleep states, modular hardware architecture, and embedded non-volatile memory ready for high-volume tapeout at a Tier 1 foundry. If it was that easy, everyone would have done it already.
You say RRAM = off the shelf where? Who supplies 16nm RRAM blocks today with proven selector integration that doesn’t kill endurance or performance? There’s a reason no supplier is named in the release, yet just four months ago at Embedded World, the 22nm demo chip showcasing that exact config was publicly tied to a company whose ReRAM is known for rad-hard use cases, edge AI, and monolithic integration.
Let’s not pretend this is some DIY kit anyone can throw together. The moat isn’t in the open-source building blocks it’s in how they’re brought together, validated, and tuned for vertical markets like defence, aerospace, and industrial AI where low power, fast wake-up, and memory resilience actually matter.
ECS-DoT isn’t just a very good product it’s a quietly strategic platform. It’s built for edge AI environments where most competitors fall apart, and it’s got the pieces and likely the partners to do it at scale. You’re looking at a parts list. I’m looking at the blueprint. Big difference.
