We started by building a multi-model AI chip on SKY130.
Then the chip taught us what company we actually needed to build.
At the beginning, WIOWIZ was not an EDA company.
We wanted to build silicon.
So we did what a small semiconductor startup does: RTL, verification, synthesis, physical design, DFT, timing, signoff preparation, pushing the design further and further toward tapeout.
Then reality started appearing, one stage at a time.
A tool was too expensive. We tried an open-source route.
A capability was missing. We built something around it.
That exposed another gap. We built again.
Simulation became a problem → we worked on simulation.
CDC/RDC needed deeper visibility → we built structural analysis.
Verification closure needed better reasoning → formal and coverage followed.
DFT became a wall → we started building the missing infrastructure.
Physical implementation exposed another set of problems → more tooling followed.
Eventually, something became obvious:
We were spending almost as much effort building the infrastructure required to build the chip as we were building the chip itself.
That changed the direction of WIOWIZ.
We postponed the tapeout.
Not because the chip journey failed.
Because the journey showed us a much larger problem worth working on.
Today we are building WIOWIZ around that lesson: an increasingly connected EDA stack, built from problems we encountered while actually trying to take silicon through the flow.
There is still a long distance between what we have and the mature commercial EDA ecosystems built over decades.
We know that.
But now we also know exactly why we are building.
The chip was the project. The problems became the product.
The full story:
https://wiowiz.com/we-started-by-trying-to-build-a-chip.html
