Optical interconnect silicon for AI infrastructure scaling
Ayar Labs builds photonic I/O chips designed to move data between AI processors faster and with less power than traditional copper interconnects. The stack—Synopsys (ICC2, Fusion Compiler), Cadence Innovus, SystemVerilog, SerDes protocols—reflects deep ASIC and silicon design work. Hiring is heavily weighted toward senior and principal engineers (40+ roles) with only 5 product positions, indicating a company still in silicon bring-up and design optimization phases, not yet in go-to-market scaling.
Notable leadership hires: Engineering Director
Ayar Labs develops optical I/O solutions for AI system interconnect, targeting the data-movement bottleneck that grows sharper as model sizes and training clusters expand. Founded in 2015 and based in San Jose, the company operates across photonics design, firmware development, and co-packaged optics integration. Active projects span silicon bring-up, SerDes firmware, ASIC design automation, and silicon photonics verification—all grounded in solving bandwidth and power constraints for next-generation AI infrastructure. The organization is currently focused on transitioning prototypes toward high-volume manufacturing.
Ayar Labs makes optical I/O interconnect chips for AI systems, designed to move data between processors at higher speeds and lower power than copper-based solutions. The technology uses silicon photonics and co-packaged optics integrated with traditional ASIC design.
Synopsys (ICC2, Fusion Compiler), Cadence Innovus, SystemVerilog, Python, Tcl, SerDes, ARM, NVIDIA, AMD, Intel, KLayout, RISC-V, and CMOS design tools. Actively adopting SerDes optimization.
Other companies in the same industry, closest in size
Ayar Labs's technology stack, projects, and hiring signals are inferred from public hiring and company data — career pages, public listings, and company web presence — then clustered and de-duplicated. Figures are estimates that refresh over time. Read our full methodology →
This is not an official vendor or customer list. It is a technology-adoption signal inferred from public data, intended for B2B research.