Ayar Labs Raises $500 Million Series E at $3.75 Billion Valuation to Scale Optical Interconnects for AI Infrastructure

Ayar Labs has secured $500 million in Series E funding, bringing its total capital raised to $870 million and valuing the company at $3.75 billion. The round was led by Neuberger Berman and included participation from ARK Invest, Insight Partners, Qatar Investment Authority, Sequoia Global Equities, and 1789 Capital, alongside strategic investors AMD, MediaTek, Alchip, and NVIDIA.

The company says the new funding will be used to expand high-volume manufacturing and testing capacity for its co-packaged optics (CPO) platform, a technology designed to address one of the most pressing challenges in AI infrastructure: how to move vast amounts of data efficiently between chips.

The Hidden Constraint in AI: Data Movement

While AI headlines often focus on faster GPUs and larger models, a growing bottleneck sits elsewhere — in the connections between chips.

Modern AI systems rely on thousands of GPUs working in parallel. These processors constantly exchange data during training and inference. The faster and more efficiently they communicate, the more powerful the overall system becomes. But most of this communication still relies on electrical signals traveling through copper wiring.

At small scales, copper works well. At hyperscale AI levels, it becomes problematic:

• Electrical signals lose strength over distance.
• Higher bandwidth requires more power.
• Increased data rates generate more heat.
• Physical copper traces consume valuable space on boards and inside packages.

As models grow into the trillions of parameters, simply adding more GPUs is no longer sufficient. The interconnect — the system that links processors together — increasingly determines overall performance and cost efficiency.

What Is Optical Interconnect — and Why Does It Matter?

Optical interconnect replaces electrical data transmission with light.

Instead of pushing electrons through copper wires, optical systems convert electrical signals into light pulses that travel through tiny optical waveguides or fiber-like structures. At the receiving end, those light signals are converted back into electrical signals for processing.

This approach offers several advantages:

• Lower power per bit transferred
• Higher bandwidth capacity
• Reduced signal degradation over distance
• Improved energy efficiency per GPU

In simple terms, light can move data faster and with less energy loss than electricity when bandwidth demands become extreme.

For AI infrastructure, that efficiency matters. Data centers are increasingly constrained by power budgets. If interconnects consume too much energy, it limits how many accelerators can be deployed within a fixed power envelope. Optical interconnect aims to relieve that pressure.

Co-Packaged Optics: Moving Light Closer to the Chip

Optical networking itself is not new — fiber optics have been used in telecommunications for decades. What is new is co-packaged optics (CPO).

Traditionally, optical modules sit at the edge of a server board or in separate pluggable transceivers. Co-packaged optics integrates the optical components directly alongside compute chips within the same package. By shortening electrical paths and moving optical conversion closer to the processor, the system reduces energy loss and latency.

At the core of Ayar Labs’ approach is its TeraPHY optical engine, designed to integrate into standard semiconductor manufacturing and packaging workflows. Rather than forcing customers to redesign entire systems, the company positions its technology as compatible with existing accelerator and switch designs.

The goal is not incremental improvement — it is enabling thousands of GPUs to function as a more tightly unified system without overwhelming power budgets.

From Unicorn Round to Scale-Up

Ayar Labs’ latest raise follows its $155 million Series D in late 2024, which pushed the company past the $1 billion valuation mark and supported early production ramp-up of its optical I/O platform.

The Series E marks a shift from validation to scale. The company plans to expand manufacturing and testing capacity, grow its presence in Taiwan’s semiconductor ecosystem, and accelerate commercial deployment.

The presence of both institutional capital and strategic semiconductor investors in this round signals that optical interconnect is no longer viewed as purely experimental. As AI infrastructure buildouts intensify, interconnect efficiency has become central to performance, cost, and energy sustainability.

If GPUs represent the engines of modern AI, optical interconnect may determine how far and how efficiently those engines can run.