Why now

The interconnect inside AI systems is growing in both volume and criticality faster than conventional copper can keep up. The ecosystem needs a way forward that preserves everything already built around it. That is the opening — and the timing — we are built for.

Principles

What guides the design

Fit the existing ecosystem

Work with the cables, connectors, modulation, and SoC silicon the industry already builds around, so adoption does not require the world to start over.

Span the full reach

Our products will span the entire spectrum of the AI datacenter — from in-package links to across-the-rack cable — covering the full 0–50m reach.

Beyond 200G, efficiently

Push per-lane bandwidth past where conventional copper runs out of room, while holding the line on reach and energy-per-bit instead of trading one for the other.

First-principles analog + DSP

Deep mixed-signal and DSP co-design — the disciplines that carried the last several generations of high-speed links — applied to the problems the next generation creates.

Production from the start

Designed for high-volume manufacturing: calibrated for consistent performance, with the diagnostics and monitoring that large AI deployments require.

Modulation- and channel-aware

Flexible and scalable across modulation schemes and channel conditions, so the same approach extends as standards and systems evolve.

Stealth mode

Detailed architecture and results are shared under NDA with partners, candidates, and investors. Reach out to go deeper.