Staera Semiconductor Launches With $4.6 Million
The San Francisco startup is developing optical technology for linking chiplets in AI accelerators.
Updated on Nov. 10, 2026 in Tech

Staera Semiconductor Inc., based in San Francisco, launched with $4.6 million in pre-seed funding led by Singaporean fund Openspace Capital. The startup develops technology for linking chiplets, beginning with its first product, Lightwire.
Why it matters
As AI accelerators demand faster data movement, copper wires used in interposers and bridges lose performance over millimeter-to-meter distances. Optical links offer a faster, lower-power alternative that can transmit data over longer distances without slowing down.
Lightwire is rated at up to 8 Gbps per link and more than 600 optical links per square millimeter, with a Lightwire-powered HBM4 memory stack projected to reach 24 Tbps against the 2.8 Tbps of current HBM4 stacks. These figures are company-reported, and independent validation has not been reported.
The players
Staera Semiconductor Inc.
A San Francisco startup developing optical interconnect technology for linking chiplets in high-performance computing and AI hardware.
Openspace Capital
A Singaporean venture fund that led the pre-seed round in Staera Semiconductor.
The details
Staera's Lightwire transmits data over optical waveguides made of silicon nitride, using light produced by miniature gallium nitride LEDs. The company plans to enhance the product, grow its U.S. team, and build out a manufacturing supply chain with the new funding.
Timeline
November 10, 2026: Staera launches with $4.6 million in pre-seed funding.
2030: Staera hopes to begin commercial shipments of Lightwire.
The Tech Race
Staera is positioning optical interconnects against the copper-based chiplet links that current HBM4 memory stacks rely on, which top out at 2.8 Tbps. Its approach targets the performance ceiling that copper hits over millimeter-to-meter distances inside AI accelerators.
If Lightwire's specifications hold, AI accelerators could move data far faster and with less power, which could eventually affect the speed and cost of cloud AI services. Commercial shipments are not expected until 2030, so near-term consumer impact is limited.
The takeaway
Optical chiplet links may be one path to overcoming copper's distance limits in AI hardware. Readers tracking AI infrastructure should watch whether Staera's performance claims are validated before commercial shipments begin.
Further reading
For more on chip interconnect developments, explore Semiconductors coverage.






