Xscape Photonics has been selected by the U.S. Department of Energy’s Advanced Research Projects Agency–Energy (ARPA-E) for support under the SCALEUP Ready program, providing funding to accelerate the commercial scale-up of its photonic technology for high-performance computing.
The Santa Clara, California-based company is developing a scalable comb laser designed for integrated photonic computing platforms. The technology is intended to address the growing energy and data-movement challenges associated with high-performance computing and AI infrastructure by enabling high-bandwidth optical connectivity between computing resources.
Under the project, Xscape Photonics plans to advance the manufacturing and commercialization of its comb laser technology. ARPA-E said the effort is aimed at enabling integrated photonic computing platforms while strengthening U.S. capabilities in advanced semiconductor and photonics technologies.
Xscape Photonics was one of two companies selected in the latest round of the SCALEUP Ready program. ARPA-E announced commitments of up to $40 million across the two projects. The other selection, Thea Energy, will establish a domestic production line for modular high-temperature superconducting magnets used in fusion energy systems.
The SCALEUP Ready program is designed to help technologies previously supported by ARPA-E move from successful research and development toward commercial-scale manufacturing and deployment. Projects are intended to demonstrate manufacturability, reliability, system integration, and performance at scales that can support subsequent private-sector investment and commercialization.
Profile: Xscape Photonics
Silicon Photonics • Multi-Wavelength Lasers • AI Data Center Interconnects
Founded
2022
Origins
Spun out of silicon photonics, nonlinear photonics, and optical interconnect research originating at Columbia University.
CEO
Vivek Raghunathan, Co-Founder and CEO; former silicon photonics product architect and program leader at Broadcom, with prior roles at Rockley Photonics and Intel.
Core Technology
Silicon PhotonicsOptical Frequency CombsMulti-Wavelength LasersDWDM Optical Links
ChromX Platform
ChromX is Xscape’s proprietary silicon photonics platform for scaling high-bandwidth optical connectivity in AI systems. It combines multi-wavelength light sources with dense wavelength-division multiplexing to increase the amount of data carried by optical links.
Multi-Wavelength Laser
An on-chip, optically pumped multi-wavelength laser source designed to generate many optical wavelengths from a compact source, enabling highly parallel optical channels for WDM-based AI interconnects.
Problem Addressed
Escape bandwidth bottlenecks between GPUs, accelerators, memory, and networking fabrics as electrical interconnects face increasing bandwidth, power, and thermal constraints.
Architecture
Uses dense wavelength-division multiplexing (DWDM) to transmit many independent optical data channels over photonic links, targeting substantially higher aggregate bandwidth density than conventional low-wavelength-count optical interfaces.
Performance Target
Xscape says its integrated optical fabrics are designed to enable petabit-per-second-scale chip-escape bandwidth with energy efficiency below one picojoule per bit.
AI Infrastructure Role
Targets optical connectivity for GPU-to-GPU, GPU-to-memory, chip-to-chip, and node-to-node communication in large-scale AI clusters and high-performance computing systems.
Products
FalconXEagleX Evaluation Kit
Manufacturing Partner
Tower Semiconductor collaborated with Xscape on prototyping and validation-kit availability for its on-chip multi-wavelength laser technology.
Strategic Investors
NVIDIACisco InvestmentsAltair
Funding
$44 million Series A announced in 2024, following earlier seed financing that included a $10 million investment from Altair.
Government Program
Selected in July 2026 for the U.S. Department of Energy ARPA-E SCALEUP Ready program to accelerate commercial scale-up of its photonic technology.
Strategic Objective
Move high-bandwidth data communication inside AI systems from increasingly constrained electrical interconnects toward massively parallel, energy-efficient optical fabrics capable of scaling future AI compute architectures.
