Powering the Future of AI with Optical Networking

Why We Invested in iPronics: Catalight Capital’s First Investment

Today marks an important milestone for Catalight Capital: our first investment.

We created Catalight to invest in the foundational technologies that will shape the next generation of computing, from semiconductors and photonics to networking and AI infrastructure.

iPronics sits directly at the intersection of those themes.

As artificial intelligence drives an unprecedented buildout of compute infrastructure, the challenge is no longer simply how many GPUs can be deployed. Increasingly, it is how efficiently those GPUs can be connected.

That shift is creating a fundamental opportunity in optical networking.

AI’s next bottleneck is the network

AI infrastructure is pushing conventional data center networks toward their limits in bandwidth, power and scale.

Today, the highest-bandwidth connections between GPUs rely heavily on copper. But as AI systems grow beyond the boundaries of a single rack, the physical limitations of copper become increasingly important and scale-up connectivity begins to move toward optics.

At the same time, conventional electrical switching requires optical signals to be converted into the electrical domain and back again. At AI scale, these conversions contribute to power consumption and complexity and can create new bottlenecks in the network.

We believe this creates an opportunity to rethink the architecture.

Why optical circuit switching

Optical circuit switching establishes direct light paths between endpoints, allowing data to remain in the optical domain rather than undergoing repeated optical-electrical-optical conversion.

AI workloads are particularly well suited to this architecture. Training and inference jobs often have communication patterns that remain relatively stable over meaningful periods of time, allowing the network to be configured around the requirements of a workload.

This has the potential to create AI infrastructure that is more flexible, more power-efficient and better able to scale.

Why iPronics

iPronics is taking a differentiated approach to this opportunity by building optical circuit switching on silicon photonics.

Its technology integrates a programmable optical switching fabric onto a chip, combining photonics with software that controls how light moves through the network.

Its flagship iPronics ONE brings this technology into a rack-ready optical networking platform with integrated control, telemetry and APIs, enabling AI clusters to reconfigure connectivity in real time.

For us, this combination is particularly compelling.

The future of AI infrastructure will not be defined by compute alone. It will require innovation across the entire system: compute, memory, interconnects, networking and photonics, working together at unprecedented scale.

We believe programmable optical networking can become an important part of that architecture.

Deep technology, built for real infrastructure

iPronics was founded in 2019 as a spin-off from Universitat Politècnica de València, building on years of research in programmable photonics.

Founder and CTO Daniel Pérez-López helped pioneer the programmable photonic technology behind the company, while CEO Christian Dupont brings extensive experience across semiconductors, communications and optical technologies.

That combination of deep technical expertise and experienced operational leadership matters enormously when moving a technology from research into commercial infrastructure.

With its new $125 million Series B, iPronics is now entering its next phase: scaling operations, accelerating commercial deployments and expanding its presence in the United States.

A first investment that reflects our thesis

As Catalight’s first investment, iPronics is also an expression of the kind of company we want to back.

We look for foundational technologies addressing important infrastructure challenges; differentiated innovation built on deep technical expertise; and teams with the ambition to build global, category-defining companies.

AI is creating enormous demand for new infrastructure, but solving that demand will require innovation far beyond the GPU.

Photonics is one of the technologies we believe will be fundamental to that transformation.

As Young Sohn said:

“Next-generation AI infrastructure will require a more dynamic, optical-first network fabric. iPronics’ approach to programmable optical switching can help data centers scale performance while managing power and cost.”

We are proud to make iPronics Catalight Capital’s first investment and to support Christian, Daniel and the entire team as they build the optical networking infrastructure for the AI era.