Show summary Hide summary
Three engineers who built laser communication systems for Starlink have spun out a Los Angeles startup and raised $50 million in a Series A round led by Thrive Capital, aiming to supply the optical hardware that next‑generation AI data centers require. Their plan targets a fast ramp to volume production of transceivers, a component that could reshape the economics and supply chains behind large GPU clusters.
Mesh Optical Technologies was founded by Travis Brashears, Cameron Ramos and Serena Grown‑Haeberli after they worked together at SpaceX, where they designed the optical links that keep thousands of satellites connected. The team says that experience exposed gaps in the market for high‑volume, cost‑efficient optical hardware.
Why optical transceivers matter now
Gallup ends long-running presidential approval tracker: analysts face new data gap
Trump claims credit for prisoner release: it predated his presidential run
Optical transceivers translate light signals from fiber or laser systems into the electrical pulses computers use. They are critical in data centers that stitch together many GPUs for model training and inference — the backbone of modern large‑scale AI.
Mesh’s founders argue the demand multiplies quickly: for every GPU added, several transceivers are required to interconnect the cluster. That scaling effect creates a large, recurring hardware need for hyperscalers.
- Funding: $50 million Series A led by Thrive Capital.
- Founders: Three former SpaceX optical engineers.
- Product focus: High‑volume optical transceivers for data centers.
- Production goal: Targeting roughly 1,000 units per day within a year to qualify for large bulk contracts in 2027–2028.
- Market context: Supply currently dominated by Chinese manufacturers; some Western suppliers already serve large cloud providers with multibillion‑dollar deals.
- Technical edge: Design changes that remove a power‑hungry component could lower GPU cluster energy use by an estimated 3–5%, according to company engineers.
Supply chains and national security considerations
Mesh positions itself as a U.S. alternative to Asia‑centric manufacturing. Investors framed the strategy partly as a response to geopolitical risk: routing critical parts of AI infrastructure through rivals or competitors could create vulnerabilities as AI investment scales.
Philip Clark, a partner at Thrive, told the press that the firm sees Mesh as addressing both a near‑term engineering bottleneck and a longer‑term supply‑chain exposure linked to AI expansion.
Manufacturing is the hard part
The founders acknowledge the toughest barrier is executing highly automated, “lights‑out” production in the U.S.—a capability that is still concentrated overseas. They point out that much of the ecosystem for high‑volume photonic manufacturing currently orients toward Chinese customers, complicating efforts by new Western entrants.
To counter that, Mesh plans to colocate design and manufacturing, a setup the company believes will reduce costs and shorten iteration cycles. Early design tweaks—most notably removing a common, power‑intensive module—are claimed to deliver measurable efficiency gains for end users.
Those savings may sound incremental, but at hyperscale they translate into significant operational cost reductions and lower power draw across entire clusters.
Bigger ambitions: photonics beyond the data center
While the immediate market is AI infrastructure, the company describes a broader vision: a gradual shift from radio frequency interconnects toward photonic communications across many device classes. The founders say they intend to begin with servers and networking hardware, then extend into other areas where high‑bandwidth, low‑latency links matter.
That roadmap depends on successfully scaling production and convincing large cloud customers to qualify a new supplier—typically a multi‑year process that requires sustained output, reliability and cost competitiveness.
For now, Mesh is betting that early traction, a manufacturing plan focused on automation, and a timely infusion of capital will position it as a viable alternative to established suppliers — a development that could influence how quickly AI infrastructure grows and where its supply chains are anchored.
