Ayar Labs Closes $500M Series E, Accelerates Volume Production of Co-Packaged Optics

Key Takeaways

• $500M Series E lifts valuation to $3.75B
• Funding drives high‑volume CPO production and testing
• Investors span chips, finance, and sovereign funds
• CPO replaces copper, boosting AI throughput per watt
• Taiwan office strengthens ASIC ecosystem partnerships

Pulse Analysis

AI workloads are hitting a power wall as traditional copper interconnects struggle to keep pace with exploding bandwidth demands. Each additional gigabit per second adds heat and consumes more energy, eroding the cost efficiency of large‑scale GPU clusters. Co‑packaged optics—integrating photonic transceivers directly with ASICs—offers a path to break this barrier by delivering terabit‑scale links with dramatically lower power per bit, a shift that data‑center operators view as essential for sustaining AI growth.

Ayar Labs’ $500 million Series E round underscores the market’s confidence in that shift. Led by Neuberger Berman and joined by a mix of semiconductor specialists (Alchip, MediaTek), venture firms (ARK, Sequoia) and sovereign investors (QIA), the round pushes the company’s valuation to $3.75 billion. The cash will fund high‑volume manufacturing of its TeraPHY optical engine, expand test capacity, and support a new Hsinchu office to deepen ties with ASIC designers. By aligning with strategic partners such as AMD and NVIDIA, Ayar positions its CPO solution for seamless integration into existing accelerator and switch ecosystems.

The broader impact extends beyond Ayar’s balance sheet. Data‑center owners stand to gain up to 30 percent lower energy costs per AI inference when copper is replaced with optical links, translating into faster ROI for hyperscale deployments. Competitors in the silicon‑photonic space must now accelerate their own production roadmaps or risk losing market share. As AI models continue to scale, the industry’s pivot toward co‑packaged optics is likely to become a defining trend, reshaping supply chains, design standards, and the economics of next‑generation compute infrastructure.