The Valencian company iPronics, specializing in the development of photonic infrastructure for artificial intelligence, has closed a Series B funding round of €108 million (approximately $125 million at the current exchange rate). The round, announced in early September, was co-led by two US venture capital firms: Maverick Silicon and Light Street Capital.
The most striking aspect of this funding round is Nvidia's entry as a strategic investor , a move that reinforces the semiconductor giant's commitment to optical networks as a key component in AI data centers. Alongside Nvidia, other new investors included Triatomic Capital, Bosch Ventures, Catalight Capital, the European Innovation Council Fund, The Tate Family Trust, Fine Structure Ventures, Amadeus Capital Partners, and Build Collective, in addition to Criteria Venture Tech, which was already a shareholder. With this injection of funds, the company's total funding reaches €153 million ($177 million).
A round that reinforces Valencia's position as a technology hub
The deal represents one of the largest Series B funding rounds ever secured by a European deep tech company and puts Valencia on the map for artificial intelligence infrastructure. iPronics was founded in 2019 as a spin-off from the Polytechnic University of Valencia , the result of more than a decade of research in programmable photonics by the iTEAM group. Since then, the company has developed a technology that replaces static network topologies with a software-controlled optical backbone.
Much of the appeal of this funding round lies in the investors behind it. Triatomic Capital is linked to Jeff Huber , former senior vice president of Google; Build Collective is led by Tony Fadell , one of the creators of the iPod and the first generations of the iPhone; and Amadeus Capital Partners was co-founded by Hermann Hauser, a key figure in the creation of ARM. Catalight Capital is also participating; its founding member, Young Sohn, was corporate president and chief strategy officer of Samsung.
Beyond the capital investment, the deal adds two top-tier professionals to iPronics' board of directors: Manish Muthal, senior managing director of Maverick Silicon , and Geoffrey Tate, former CEO of Rambus and co-founder of FlexLogix (acquired by Analog Devices). Tate, who also sits on the board of Ayar Labs, brings decades of experience in technology leadership, first as senior vice president of microprocessors at AMD. Young Sohn joins as an advisor to the company.
What is iPronics ONE and why is it of interest to Nvidia?
iPronics' flagship product is iPronics ONE (Optical Networking Engine) , a rack-mounted optical circuit switch that enables dynamic reconfiguration of connectivity between GPU clusters in real time. Unlike a traditional switch that routes packets electronically, this device establishes direct optical paths between endpoints and can modify those connections in less than 300 microseconds, with latency of less than 30 nanoseconds.
The company's argument is that current AI clusters force users to choose a fixed network topology and live with it. As a cluster grows, this rigidity translates into idle GPUs, extra power consumption, and costly reorganizations . iPronics' platform allows for on-demand network rewiring, adapting to training or inference workloads and even diverting traffic when a component fails. According to company data reported by Reuters, the hardware reduces the size of current OCS equipment by about 20 times, a key metric in installations where every rack unit counts.
Nvidia's interest is no coincidence. Jensen Huang's company understands that every network improvement that increases the utilization of its accelerators makes its entire platform more valuable . If a programmable optical switch allows a GPU to spend less time waiting for data, the total cost of ownership of the cluster decreases, and the decision to purchase more GPUs becomes easier for the customer. Furthermore, Nvidia's involvement lends legitimacy to iPronics in discussions with hyperscalers and AI equipment manufacturers.
A market context that favors optical switching
Optical circuit switching (OCS) is not a new technology. Google has been deploying it in its own data centers for years to connect its AI supercomputers, but the market is now opening up to more players as clusters scale from a single rack to tens of thousands of accelerators. Networking has become as critical a factor as computing power, and OCS is positioning itself as an alternative to traditional copper-based architectures and electronic packet switching.
iPronics founder and CTO Daniel Pérez-López explained to Reuters that AI cluster failures "occur on a timescale of minutes ," and that the primary motivation for adopting OCS is resilience: a robust network capable of self-reconfiguration. The company claims its technology can reduce required switching power by up to 50% by decreasing the number of transceivers needed. Furthermore, the integration of control, telemetry, and APIs allows operators to manage the optical layer directly from the cluster management system.
International growth and expansion plans
With this funding round, iPronics plans to double its workforce, which currently numbers around 100 employees , over the next year. It has also opened an office in Santa Clara, California, to strengthen its presence in the United States and is actively recruiting for product strategy, customer implementation, and AI infrastructure partnerships.
The startup has also established a dedicated manufacturing line with the Thai manufacturer Fabrinet, geared towards high-volume production—a crucial step in moving from the prototype phase to large-scale commercialization. According to the company, it already has a number of hyperscale and OEM clients for AI clusters, which the firm itself estimated at "the first ten" at the time of the collaboration's announcement.
In the words of Christian Dupont, CEO of iPronics: “Having the support of Maverick Silicon, Light Street Capital, and Nvidia is a powerful validation of our vision .” The CEO adds that, thanks to this investment, the company can “significantly accelerate its time to market and ensure that data centers can get the most out of their GPUs without compromising performance.” Manish Muthal, from Maverick Silicon, emphasizes that “the exponential growth of AI workloads has pushed traditional network infrastructure to the limits of its power and bandwidth,” making optical circuit switching “more crucial than ever.”
iPronics has positioned itself in a technical niche that, until recently, remained outside the headlines but is becoming increasingly strategic as clusters grow. This acquisition demonstrates that Europe can participate in the AI infrastructure sector from within technical niches , not by competing with the giants of the modeling industry, but by contributing the critical component that the global ecosystem needs: the interconnection layer that enables thousands of GPUs to work together efficiently.