Three former SpaceX engineers have left Starlink’s orbit to launch Mesh Optical Technologies, a Los Angeles startup aiming to reinvent optical transceivers for data centers. With a $50 million Series A led by Thrive Capital, they say their new architecture can cut energy consumption in GPU clusters by 3–5%—a pretty big deal for hyperscale operators facing constant throughput and cost headaches.
This post digs into who these founders are, what they’re actually changing in optical hardware, and how they plan to scale production in a supply chain landscape that’s, well, not exactly boring these days.
Mesh Optical: a new frontier in data-center optics
Mesh Optical Technologies is tackling a critical choke point in the digital infrastructure that powers today’s AI and cloud workloads. They’re rethinking the design of optical transceivers—the gadgets that switch between optical and electrical signals.
Mesh believes current designs are energy-hungry and tough to manufacture at the scale hyperscale data centers need. Their approach, according to the co-founders, leads to higher data throughput with less power draw. That’s a win for operators watching their electric bills and thermal budgets balloon.
Founders with SpaceX pedigree
The company was started by Travis Brashears, Cameron Ramos, and Serena Grown-Haeberli. These three engineers worked on Starlink’s optical inter-satellite links before deciding to build something for earthbound networks.
Their experience with high-performance optical networks shapes Mesh’s belief that a simpler, more efficient transceiver can make a real difference across data centers and the broader communications stack. After SpaceX, they set up shop in Los Angeles to chase a hardware architecture rethink they see as overdue for scaling AI workloads.
Energy savings and architecture redesign
Mesh’s big claim is that they’ve removed a power-hungry component from standard transceiver designs. By streamlining the signal path, they say their transceivers could shave about 3–5% off total GPU-cluster energy use.
For hyperscale operators, even a few percentage points of savings mean less electricity, lower cooling needs, and real cost reductions—especially as AI models keep growing.
Strategic manufacturing shift: a China-independent supply chain
The founders point out that most optical transceivers come from Chinese suppliers. With geopolitics and national security on everyone’s mind, Mesh wants to build a China-independent supply chain while still hitting tough cost and quality targets.
They’re hoping to tighten the feedback loop between design and production and dodge the risks of relying on a single country for key components.
Co-located design and production
Mesh plans to bring design and manufacturing together in the U.S. They’re aiming for fully automated, “lights-out” manufacturing—no small feat, since so much expertise is still clustered in China and some suppliers prioritize their local customers.
By keeping everything under one roof, Mesh hopes to speed up iteration, boost yield, and bring down costs as they ramp up production.
Scaling toward mass production
The short-term goal? Hit 1,000 units per day within a year. That’s the threshold to get into bulk qualification cycles expected in 2027 and 2028.
If they pull it off, Mesh could go toe-to-toe with bigger players during industry qualification rounds, and be ready to support the fast-moving world of AI and data-center rollouts.
Funding, validation, and market outlook
Mesh’s $50 million Series A, led by Thrive Capital, shows investors are betting on their hardware-first approach to data-center efficiency. The market is wide open: while they’re starting with optical transceivers for GPU clusters, the team sees a future in other sectors that need high-speed, energy-efficient data transfer.
Investors and strategic implications
With Thrive Capital’s backing, Mesh gets more than money—they tap into a strategic partner network to help navigate early manufacturing and qualification hurdles. Their focus on a U.S.-based, automated model could attract more investors interested in domestic semiconductor and electronics supply chains.
That’s a goal that lines up well with national-security priorities and the bigger push for supply chain resilience.
Beyond data centers: a broader application horizon
Right now, the main goal is to cut energy use in GPU-heavy data centers. The founders seem pretty hopeful about bringing their wavelength-optical know-how to other fields that need high-speed, efficient data transfer.
If Mesh actually hits its targets for manufacturing and performance, we might see this tech show up in AI accelerators, HPC clusters, and enterprise networks that want cleaner, smarter optical interconnects. That kind of expansion could shake up efficiency standards all over the digital infrastructure world.
Here is the source article for this story: Ex-SpaceX engineers grab $50M for Mesh Optical to power AI clusters