Mojo Vision, a start-up based in Cupertino, California, announced a $17.5 million strategic investment from Future Ventures to accelerate commercialization of its micro-LED platform for next-generation optical I/O in AI infrastructure. The Cupertino-based company said the funding follows its $75 million financing in August and will support continued development of its wafers-in, wafers-out micro-LED technology designed for high-bandwidth, energy-efficient interconnects.
The company is targeting a growing bottleneck in AI data centers, where bandwidth density and power efficiency increasingly constrain system scaling. Mojo Vision’s architecture replaces traditional laser-based optics with dense arrays of micro-LED emitters, enabling massively parallel optical links that increase bandwidth density while reducing energy per bit.
The announcement builds on Mojo Vision’s recently disclosed collaboration with Marvell to develop high-density micro-LED connectivity solutions for AI data center environments. The companies aim to address short-reach interconnect requirements in hyperscale systems as AI workloads drive demand for new connectivity architectures beyond copper and conventional optical links.
- $17.5 million strategic investment led by Future Ventures, the VC firm co-founded by Steve Jurvetson
- Follows Mojo Vision’s $75 million financing round in August 2025
- Focus on commercialization of micro-LED-based optical I/O for AI infrastructure
- Architecture uses GaN-on-silicon micro-LED emitters, silicon photodetectors, multicore fiber, and micro-lens arrays
- Designed to enable thousands of optical channels in a compact footprint with lower energy per bit
- Strategic collaboration with Marvell targets high-density optical interconnects for AI data centers
- Technology applicable to terrestrial AI data centers and emerging distributed or orbital compute environments
“AI infrastructure is reaching fundamental limits in bandwidth density and power efficiency, and incremental improvements are no longer enough,” said Nikhil Balram, CEO of Mojo Vision. “Our micro-LED platform was purpose-built to overcome this tradeoff, enabling thousands of optical lanes in a compact footprint and unlocking major gains in bandwidth while lowering energy per bit. This investment accelerates our path to bringing a new class of optical interconnect solutions to market.”
Mojo Vision — Company Snapshot
| Headquarters | Cupertino, California, USA |
| Founded | 2015 |
| Founders | Drew Perkins, Mike Wiemer, Brad Bell, Joe Schellenberg |
| CEO | Nikhil Balram |
| Total Funding Raised (since 2023 pivot) | $136M |
| Latest Investors | Future Ventures (Steve Jurvetson) |
| Technology Focus | Micro-LED-based optical I/O for AI infrastructure |
| Core Architecture | GaN-on-silicon micro-LED emitters, silicon photodetectors, multicore fiber, micro-lens arrays, CMOS control circuitry |
| Key Value Proposition | Massively parallel optical interconnects enabling higher bandwidth density and lower energy per bit |
| Primary Use Cases | AI data center interconnects, scale-up fabrics, short-reach optical links, emerging distributed compute systems |
| Key Products / Platform | Micro-LED optical I/O platform (wafers-in, wafers-out integrated solution) |
| Strategic Partnerships | Marvell (micro-LED optical interconnect solutions for AI infrastructure) |
| Notable Differentiation | Replaces laser-based optics with dense micro-LED arrays to enable thousands of parallel optical channels in a compact footprint |
🌐 Analysis: The micro-LED optical interconnect segment is rapidly evolving into a distinct category within AI infrastructure, with multiple startups and ecosystem players advancing parallel approaches to short-reach, high-density optical I/O. Mojo Vision’s architecture—based on massively parallel arrays of emitters—aligns with a broader “slow-and-wide” design philosophy aimed at reducing power while scaling bandwidth inside AI clusters.
Avicena has emerged as one of the most visible competitors, focusing on its LightBundle micro-LED interconnect architecture for scale-up networks. The Sunnyvale-based company raised a $65 million Series B in 2025 with backing from SK hynix and introduced an evaluation kit in March 2026, signaling a transition from concept to system-level validation. Avicena has also worked with TSMC on silicon photodetector arrays, indicating early supply chain alignment.
Credo entered the space through its planned acquisition of Hyperlume, a startup founded in 2022 that developed micro-LED-based chip-to-chip optical interconnects. Hyperlume raised a $12.5 million seed round in early 2025 before the deal. Credo’s move positions micro-LED as a potential extension of its connectivity portfolio for AI scale-up fabrics, complementing its established DSP and retimer business.
Marvell, while not a micro-LED developer, is becoming a key ecosystem integrator. Its collaboration with Mojo Vision suggests micro-LED optical I/O could integrate with broader AI interconnect architectures that include optical DSPs, co-packaged optics, and switch silicon. Additional ecosystem signals include Microsoft Research and MediaTek demonstrating micro-LED-based active optical cables, and ams OSRAM and CEA-Leti advancing component and system-level programs. Together, these efforts indicate that micro-LED optical I/O is moving from research into early commercialization, with multiple architectural paths now competing to address power and bandwidth constraints in AI infrastructure.
🌐 We’re tracking the latest developments in networking silicon. Follow our ongoing coverage at: https://convergedigest.com/category/semiconductors/
🌐 We’re launching the “Data Center Networking for AI” series on NextGenInfra.io and inviting companies building real solutions—silicon, optics, fabrics, switches, software, orchestration—to share their views on video and in our expert report. To get involved, send a note to [email protected] or [email protected].
