SpaceX vets raise $50M to mass-produce American-made optical transceivers for AI data centers

SpaceX vets raise $50M Series A for data center links | TechCrunch

Travis Brashears, Cameron Robinson and Serena Grown-Haeberli began collaborating at SpaceX, developing optical communications links that keep thousands of Starlink internet satellites in constant contact. Now, the three engineers are co-founders of Mesh Optical Technologies, a Los Angeles startup that announced a $50 million Series A led by Thrive Capital on Tuesday. Mesh aims to mass-produce optical transceivers, devices that convert optical signals from fiber or laser into electrical signals for computers. CEO Brashears, President Robinson, and VP of Product Grown-Haeberli realized the opportunity when designing a new generation of compute-hungry SpaceX satellites forced them to assess the optical transceiver market, and they saw its limitations. Optical transceivers are particularly important for data centers aimed at training and operating large deep learning models, because they allow multiple GPUs to work in concert. One established U.S. supplier, AOI, won a contract worth $4 billion to provide components for AWS data centers last year. "Someone will brag about a million GPU cluster; you have to multiply by four to five for the number of transceivers in that cluster," Brashears explained. The company's goal is to manufacture a thousand units per day within the year so they can begin qualifying for bulk orders in 2027 and 2028. The optical transceiver market is dominated by Chinese firms and suppliers, and Mesh sees an advantage in building its supply chain outside of that country. While trade restrictions haven't impacted the market yet, the founders and their backers see themselves as getting in front of a national security dilemma. "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," Thrive Partner Philip Clark wrote TechCrunch. "In the immediate term, Mesh is solving our need for better ways to do interconnect if we want to keep scaling AI." The challenge for Mesh, the founders say, is executing lights-out, automated manufacturing techniques, which aren't common in U.S. industry. So much of this expertise is concentrated in China that even European equipment suppliers expect Chinese customers -- one German firm's standard intake form asks for a Chinese company registration number. By co-locating design and production, the founders hope to realize more efficient and lower-cost components. Their current design removes one commonly used but power-hungry component, which Robinson said could reduce GPU cluster power usage by 3% to 5%, a meaningful amount as hyperscalers seek to wring as much efficiency out of their systems as possible. Data centers are just the beginning of Mesh's aspirations; the company sees optical wavelength communications as the next paradigm in communications. "The world has primarily focused on [radio frequencies] for a long time," Brashears told TechCrunch. "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."

Mesh Optical Technologies raises $50M to mass produce American-made data center links - SiliconANGLE

Mesh Optical Technologies raises $50M to mass produce American-made data center links Three SpaceX Corp. alumni who helped the rocket company develop optical communications for its Starlink satellite Wi-Fi service have raised millions of dollars to commercialize that technology for data centers. Mesh Optical Technologies, co-founded by its Chief Executive Travis Brashears, President Cameron Ramos and Vice President of Product Serena Grown-Haeberli, said today it has raised $50 million in Series A funding. The round was led by Thrive Capital and saw participation from Also Capital and Banner VC, Bloomberg reported. The company's valuation was not disclosed. The startup's goal is to mass-produce high-performance optical transceivers, which are devices designed to move data around data centers at rapid speeds. They work by converting optical signals from fiber or lasers into electrical signals that can be read by computers. The co-founders told Bloomberg that they first realized the opportunity for this technology while working at SpaceX. Tasked with finding a way for Starlink's satellites to communicate vast amounts of data in real time, they began looking at optical receivers, and quickly realized the industry's limitations. Optical receivers have become an essential element of modern artificial intelligence data centers, also known as "AI factories," which support the training and inference operations of large language models. They make it possible for multiple graphics processing units and other chips to work together in unison, so AI companies can build the massive clusters their largest models require. Traditional data center architectures don't ill-suited for GPU clusters, because they were not designed for the kind of lateral, high-throughput traffic that characterizes today's AI factories. Historically, cloud data centers have always been optimized for "north-south" traffic flows, meaning data that moves into and out of the facility. But AI training and inference generates massive "east-west" traffic, as the GPUs are constantly bouncing data around between themselves so they can work together to tackle problems. Existing network switches, cables and interconnects were never designed to handle this kind of traffic, which is why bandwidth-busting optical transceivers have become so important. In an interview with TechCrunch, Brashears revealed that the average AI factory typically uses millions of the devices. "Someone will brag about a million GPU cluster," he said. "But you have to multiply that by four or five to get the number of transceivers in that cluster." Optical transceivers are not new, but Mesh believes there's an opportunity because the vast majority of the world's devices are currently manufactured in China. The co-founders believe that's almost certainly going to be problematic in future. Mesh proudly proclaims that its products are "made in the USA," and that's probably going to be extremely advantageous. While the transceiver market has not yet seen any trade restrictions, the GPUs and other powerful processors that fuel the AI industry certainly have. Brashears said his company is trying to get ahead of a potential national security problem. Thrive Capital Partner Philip Clark agrees. He told TechCrunch that U.S. companies can't take the risk of having critical parts of the AI supply chain controlled by misaligned and competitive companies. Mesh's immediate goal is to scale up its manufacturing operation. It aims to produce 1,000 units per day by the end of the year, with a view to taking bulk orders for its products in 2027. Doing so will be challenging, because the kind of automated manufacturing techniques employed by Chinese factories aren't common in the U.S. Still, Ramos told Bloomberg that the company aims to set itself apart by making its optical transceivers more efficient than their Chinese counterparts. Crucially, his team has designed a transceiver that removes one of the most power-hungry components, allowing it to improve its energy efficiency by between 3% and 5%. That might not seem like a lot, but for hyperscale AI factories packed with hundreds of thousands of GPUs, it could translate to significant savings on their energy costs. Mesh's next step is to move beyond prototyping and testing to commercial production. At the same time, it's going to engage with prospective customers, including large-scale data center operators and cloud infrastructure providers, in order to validate its technology's integration and performance.