SpaceX Vets Raise $50M for Data Center Optics

What did Mesh Optical Technologies just announce?

Mesh Optical Technologies raised $50 million in a Series A round led by Thrive Capital. The Los Angeles startup announced the funding on February 17, 2026. Also Capital and Banner VC also participated in the round, according to Bloomberg. The company's valuation was not disclosed.

Who founded Mesh Optical Technologies?

The three co-founders — Travis Brashears, Cameron Ramos, and Serena Grown-Haeberli — all met at SpaceX. At SpaceX, they developed optical communications links for the Starlink satellite internet service. Brashears is CEO, Ramos is President, and Grown-Haeberli is VP of Product.

The founders said they spotted the opportunity while designing a new generation of compute-heavy SpaceX satellites. That work forced them to assess the optical transceiver market closely, and they saw its limits firsthand.

What is an optical transceiver?

An optical transceiver is a device that converts optical signals from fiber or laser into electrical signals that computers can read. These components are critical for connecting large numbers of GPUs inside AI data centers, allowing them to work together on training and running large deep learning models.

As Brashears put it to TechCrunch: "Someone will brag about a million GPU cluster; you have to multiply by four to five for the number of transceivers in that cluster."

Why does the optical transceiver market matter for AI?

Traditional data center designs were built for "north-south" traffic — data moving in and out of a facility. AI factories need high-throughput lateral traffic between thousands of GPUs. Optical transceivers make that kind of interconnect possible at scale.

You might also like

Google Expands Publisher AI Licensing Program

Micron AI Memory Supercycle May Last Until 2030

Oracle Cuts 21,000 Jobs to Fund AI Expansion

Apple Vision Pro Chief Joins OpenAI Hardware Team

One established U.S. supplier, AOI, won a $4 billion contract to supply components for AWS data centers. That gives a sense of how large the market has become as hyperscalers build out AI infrastructure.

Here's a quick look at the key facts from the Mesh raise:

Why is Mesh building in the U.S. instead of China?

The optical transceiver market is dominated by Chinese firms. Mesh sees a strategic opening in building its supply chain outside China. The founders and their backers believe trade restrictions could eventually hit the market, and they want to be ahead of that.

Thrive Partner Philip Clark wrote to TechCrunch: "If AI is the most important technology in several generations (which we believe to be true), to have critical parts of AI data center capex run through misaligned/competitive countries is a problem."

The challenge is real. Lights-out, automated manufacturing expertise is concentrated in China. Clark noted that even European equipment suppliers default to Chinese customers — one German firm's standard intake form asks for a Chinese company registration number.

This dynamic echoes broader supply chain concerns we've tracked across the AI hardware stack, from Nvidia banned chips to ASML EUV export worries.

What does Mesh's design actually improve?

Mesh's current transceiver design removes one commonly used but power-hungry component. Ramos said this could cut GPU cluster power usage by 3% to 5%. For hyperscalers running massive clusters around the clock, that is a meaningful efficiency gain.

By co-locating design and production, the founders also expect to produce more efficient, lower-cost components than competitors who separate those functions.

What are Mesh's production targets?

The company's goal is to manufacture 1,000 units per day within the year. That production rate would let Mesh begin qualifying for bulk orders in 2027 and 2028.

Data centers are the starting point, not the endpoint. Brashears told TechCrunch: "The world has primarily focused on [radio frequencies] for a long time. We want to be at the precipice of transition from RF to photonics...we want to interconnect everything, and not just computers, but that's where we're starting."

The Starlink-to-data-center pipeline is becoming a real talent funnel for AI infrastructure. The SpaceX compute deal with Reflection AI is another sign of how deep that connection runs. As AI clusters grow, the demand for AI supercomputer infrastructure is pulling in engineers from aerospace, defense, and satellite programs.

As we see it, the Mesh story is really about where the bottlenecks in AI scaling actually sit — not just in chips, but in the unglamorous hardware that connects them. The Samsung HBM4 supply ramp is one piece; optical interconnects are another.

Mesh's next concrete milestone is reaching 1,000 units per day in production, with bulk order qualification targeted for 2027 and 2028.