LOS ALTOS — Majestic Labs, an AI startup founded by veteran chip designers from Meta and Google, emerged from stealth mode Tuesday after raising $100 million to develop technology it claims will drastically reduce the cost and scale of data centers needed for advanced artificial intelligence.
The company, with dual headquarters in Los Altos, California, and Tel Aviv, Israel, said its patent-pending server architecture can collapse the equivalent of up to 10 conventional server racks into a single unit. The founders told CNBC this is achieved by providing 1,000 times the memory capacity of a typical enterprise server, addressing a critical bottleneck in processing large AI models.
“We’re not trying to replace GPUs across the board — we’re solving for memory-intensive AI workloads where the fixed compute-to-memory ratio becomes a constraint,” said CEO Ofer Shacham, referring to the graphics processing units from industry leader Nvidia that dominate AI computing.
The funding, closed in a previously unannounced $71 million Series A round in September led by Bow Wave Capital, comes as tech giants like Alphabet, Meta, Microsoft and Amazon collectively project capital expenditures on data centers to exceed $380 billion this year. A prior seed round brought the total funding to $100 million.
Majestic’s technology is aimed at “hyperscalers” and large companies in sectors like finance and pharmaceuticals that run complex AI models. The company claims its servers will have a smaller physical footprint and require less power and cooling, directly targeting rising infrastructure costs.
The co-founders — Shacham, President Sha Rabii, and COO Masumi Reynders — first worked together at Google, where they helped build the team for the company’s custom Tensor Processing Units (TPUs). They later moved to Meta to lead its Facebook Agile Silicon Team (FAST) before founding Majestic in late 2023.
Prototypes of Majestic’s servers are expected to be available to select customers in 2027, and the startup has begun discussions about pre-orders. The company, which currently has fewer than 50 employees, declined to name its clients.
While Nvidia remains the dominant force in AI hardware, Majestic enters a growing field of competitors, including Google with its latest Ironwood TPU and other startups like SambaNova and Cerebras, all seeking to capitalize on the exploding demand for efficient AI computing power.