PicoJool raises $27.5M Series A to push VCSEL optical links into AI clusters
PicoJool, backed by former Intel CEO Pat Gelsinger, raised $27.5M to commercialize optical interconnects for AI data centers.
The Series A round was led by Socratic Partners, with Hudson River Trading also participating. The new capital brings PicoJool's total funding to $39.5 million, following a $12 million seed round last November that was led by Gelsinger's venture capital firm, Playground Global.
The company is developing vertical cavity surface emitting lasers, or VCSELs, a type of semiconductor laser diode that carries data as light signals through optical cables rather than traditional copper wire. PicoJool has built optical chips and modules around the technology and says they can improve both the performance and the energy efficiency of large AI chip clusters.
Demand for that kind of connectivity is rising because more capable AI models require larger clusters of graphics processing units and other accelerators. The largest enterprises now run clusters holding hundreds or even thousands of AI processors that constantly exchange data packets so they can work in unison on AI workloads. That traffic is straining the copper interconnects linking the chips, which PicoJool says cannot keep pace with the bandwidth demands of the biggest clusters and consume large amounts of power, raising the cost of running AI.
PicoJool founder and Chief Executive Al Yuen says the company can address the bottleneck by replacing older interconnects with its 100G and 200G VCSEL products. It also offers low-power MicroVCSEL configurations, including 50G NRZ and 64G NRZ/PAM4 interconnects. According to PicoJool, the MicroVCSEL platform uses a massively parallel architecture that aggregates bandwidth while cutting the energy cost of moving data around AI clusters, and it offers a path to connectivity of up to 3.2 terabits per second.
“Reaching 200G per lane proved that the performance is there,” Yuen said. “Our focus is now execution. This funding allows us to expand our teams in the U.S. and Taiwan, qualify products with customers and move a complete portfolio from chip-level VCSELs and MicroVCSELs into AOC and NPO modules.”
Yuen said part of the money will go toward accelerating research and development and expanding the company's operations and sales teams, though the main priority is commercializing existing technology. PicoJool is working with WIN Semiconductors Corp. and other chip foundries to prepare the 200G VCSEL for mass production, with its other products to follow.
Gelsinger said he was drawn to PicoJool by its ambition and by what he described as a workable answer to the AI data center bandwidth bottleneck that has constrained AI deployments globally. “The network is the AI, and connectivity is becoming one of the defining constraints on performance,” he said. “The industry needs a solution that can increase bandwidth while sharply improving power and cost.”