Ayar Labs, Inc.

Ayar Labs designs optical I/O chiplets (TeraPHY) and an external multi-wavelength laser light source (SuperNova) that are co-packaged alongside compute dies to replace copper interconnects with optical links for chip-to-chip and rack-scale communication in AI data centers.

Ayar Labs has moved from a DARPA-funded research spinout to a company with strategic capital and design engagement from all three major compute-silicon players — Nvidia, AMD and Intel each invested in its December 2024 Series D, and Nvidia and AMD returned as strategic investors in its March 2026 Series E, which also drew Qatar Investment Authority, MediaTek and Alchip Technologies. The company has paired that capital with concrete integration work: a November 2025 collaboration with Global Unichip Corp to build reference XPU designs using Ayar's optical chiplets, and a March 2026 partnership with server maker Wiwynn to build rack-scale, liquid-cooled AI systems around its CPO technology. It also holds a DoD prototype contract aimed at building a domestic co-packaged-optics supply chain. The Series E proceeds are earmarked explicitly for scaling high-volume production and test capacity — the step that determines whether Ayar's technology reaches real accelerator racks rather than remaining a demo. The thesis is that as accelerator interconnect bandwidth needs outrun copper, Ayar is one of the best-positioned independent optical I/O suppliers, provided it converts its multi-vendor backing into qualified, shipping volume.

Moving from roughly 15,000 shipped devices as of late 2024 to true high-volume production is an unproven manufacturing scale-up. No named hyperscaler production commitment has been publicly disclosed as of September 2026; reference-design partnerships are not the same as shipped volume. Large incumbents, including Nvidia's own optics roadmap, Broadcom and TSMC COUPE, could internalize co-packaged optics rather than buy from an independent supplier — notable since Nvidia is simultaneously an investor and a potential competitor. Co-packaged optics requires continued heavy capex for test and production capacity despite $870M raised to date; further large raises may be needed before profitability.

A hyperscaler or XPU vendor publicly cancels or de-scopes a co-packaged optics integration with Ayar in favor of an in-house or competitor solution. Ayar misses its stated volume-production timeline without a credible replacement milestone. A strategic investor divests or stops participating in future rounds, signaling reduced conviction.

Ayar Labs, Inc. appears here as a research record. Coverage does not disclose or establish investment availability, relationships, permission or allocation.