HyperLight Raises $80M to Scale TFLN Photonics for AI Infrastructure
The Cambridge-based firm will use MediaTek-led funding to deploy high-bandwidth optical interconnects for next-generation data centers.
Cambridge-based photonics firm HyperLight has raised $80 million in a Series C funding round to accelerate the deployment of its thin-film lithium niobate (TFLN) Chiplet Platform. The investment targets the critical need for higher bandwidth and lower power consumption in AI infrastructure and data centers.
The funding round was led by MediaTek and included participation from UMC Capital, Jabil, Foxconn, EDBI, CDIB-TEN Capital, Qatar Investment Authority, Summit Partners, The Engine, Foothill Ventures, and Xora Innovation. HyperLight intends to use the capital to scale its TFLN technology, which currently supports 200G-per-lane operation, while 400G-per-lane solutions are already sampling in the market.
The Shift to TFLN
As AI networks evolve, the industry faces a bottleneck in optical interconnects. Traditional solutions struggle to maintain efficiency as lane speeds increase, creating a demand for integrated photonic circuits that can scale without prohibitive power costs. HyperLight’s TFLN technology addresses this by providing a more efficient architecture for AI data centers, telecom, and metro networks.
To move these components from the lab to the data center, HyperLight has established strategic manufacturing partnerships with UMC and Wavetek. These agreements enable high-volume production using 6-inch and 8-inch wafers, providing the industrial capacity necessary to meet the demands of global AI infrastructure.
Industry Implications
This transition is essential as the industry moves toward 3.2T connectivity and beyond. By aligning with semiconductor giants like MediaTek and manufacturing leaders like UMC, HyperLight is positioning TFLN as a production-ready standard. This alignment aims to reduce ecosystem complexity and lower the overall cost of adopting high-speed optical interconnects.
"This financing is about more than capital — it is about ecosystem alignment," said Mian Zhang, CEO of HyperLight. Zhang noted that AI infrastructure requires optical interconnects capable of delivering higher bandwidth and lower power across both pluggable and co-packaged optics. Brian Hsu, Managing Director of the MediaTek Innovation Fund, added that TFLN’s efficiency makes it a compelling technology for the transition to 3.2T connectivity.
Future Outlook
Market observers will now watch for the full commercial rollout of the 400G-per-lane solutions and the speed at which the UMC and Wavetek partnerships can scale production. As AI clusters grow in size and complexity, the ability to standardize TFLN architecture will be a key determinant in whether the industry can sustain its current trajectory of bandwidth expansion.