This blog post takes a look at Ayar Labs’ latest funding milestone. It also digs into how its co-packaged optics (CPO) technology—and its TeraPHY processing chip—could shake up data center bandwidth and AI workloads.
The late-stage Series E round, led by Neuberger Berman, included strategic participation from Nvidia, AMD, and MediaTek. This signals growing financial and strategic support for optical interconnects, especially as copper links start falling behind expanding data demands.
Funding momentum for optical interconnects
Ayar Labs just landed $500 million in a late-stage Series E, which brings total outside funding to about $870 million. Neuberger Berman led the round, with big names like Nvidia, AMD, and MediaTek also investing.
This kind of support shows a wider industry trend: data centers need higher-bandwidth, lower-power interconnects as they scale up for AI and latency-sensitive tasks. The new funding will help Ayar ramp up CPO production, sharpen its testing processes, and expand into new regions.
Investors clearly see optical interconnects as a key enabler for hyperscale compute and accelerators. Copper-based links just can’t keep up anymore, especially when it comes to throughput and energy use.
Nvidia, in particular, has been making big moves, including recent multi-billion-dollar investments in optical component suppliers. With such high-profile backers and a strong technical case for light-based links, CPO technologies look set to keep building momentum as data centers chase more bandwidth with less power.
What is Co-Packaged Optics (CPO) and why it matters
Co-packaged optics puts optical components right next to a processor or accelerator, ditching old-school copper interconnects. By sending data as light, CPO can seriously boost bandwidth per watt and lower latency—huge wins for AI inference, big training jobs, and real-time analytics.
Ayar’s CPO platform pairs a flat light emitter, SuperNova, with a dedicated processing chip, TeraPHY. This setup encodes and transmits data using laser beams and integrated optical devices.
Inside TeraPHY: architecture enabling terabit-scale data transfers
TeraPHY uses millions of transistors and microring resonators to control light and create high-throughput data paths. According to Ayar, a single TeraPHY can move up to 8 terabits per second.
It also delivers lower latency by skipping traditional forward error correction in some cases. There’s a UCIe-compatible version too, which lets it plug directly into processors like GPUs. That fits right in with the interests of major AI accelerators and their partners.
Strategic investments and market momentum
As Ayar pushes its CPO strategy forward, the involvement of Nvidia and other silicon giants shows the ecosystem is shifting toward optical interconnects. Nvidia’s broader investments in optical components make it clear: next-gen data centers need light-based links to keep up with bandwidth-hungry workloads, especially real-time AI inference and large-scale model training.
The collaboration between Ayar, Nvidia, AMD, and MediaTek shows chip designers and interconnect innovators are increasingly on the same page, aiming to streamline integration and speed up deployment. It’s a fascinating moment for the industry—one that feels like it could shape the direction of data center technology for years to come.
Impact on data centers and AI workloads
The funding outcome and technical roadmap put Ayar’s CPO approach right in the middle of the data center evolution. Co-packaged optics look set to replace copper links between chips, which could mean some big changes:
- Significantly higher bandwidth per interconnect, supporting multi-terabit data paths for AI workloads.
- Lower power consumption and simpler signaling, making data centers a bit greener.
- Faster reach and lower latency for interconnects inside racks and across chips in multi-chip modules.
- Easier integration with GPUs and accelerators through standards like UCIe, which helps cut down system-level integration headaches.
Looking ahead, Ayar plans to ramp up production capacity. They’re also working on refining test workflows and growing their international presence.
The combination of strong funding and a clear technical direction hints that optical interconnects will show up more and more in AI systems, data centers, and cloud infrastructure as demand keeps rising.
Here is the source article for this story: Co-packaged optics startup Ayar Labs raises $500M round backed by Nvidia, AMD