TeRAM has emerged from stealth with $37 million in equity financing from early-stage investors, aiming to address the memory bottleneck facing AI deployment. The company is developing custom 3D SRAM technology for frontier AI applications, initially targeting data centers. As AI models scale beyond two trillion parameters, new memory technology is needed to increase capacity and throughput while reducing power consumption. TeRAM is currently validating its technology and targeting initial customer production in 2029.
Brian Schechter, a partner at Primary Venture Partners and lead investor in the seed round, said memory is the constraint on AI progress, and that incremental improvements won't be enough to solve it. He said TeRAM's CEO recognized this early, describing him as a rare founder who connects AI architecture, memory physics, and semiconductor manufacturing, and has assembled an exceptional team to realize his vision, calling TeRAM the memory platform the next generation of AI requires.
Under what's often called the AI scaling law, frontier AI models become more capable as their parameter counts and context windows grow. Combined with the rise of agentic AI workflows, this has accelerated broader AI adoption. Since mid-2025, agentic inference has overtaken model training as the primary driver of AI compute and token generation. This shift has moved the "memory wall," the point at which memory becomes the limiting factor, from high-bandwidth memory (HBM) capacity during training toward memory bandwidth during inference instead. Generating each output token requires passing through an entire model and its trillions of parameters, making memory bandwidth a bottleneck for AI compute regardless of whether the model sits in on-chip SRAM or off-chip HBM modules.
TeRAM is developing memory and packaging technology to integrate 3D SRAM directly onto AI compute chips. Designed to withstand demanding thermal conditions, this technology aims to deliver unprecedented memory bandwidth at the lowest possible power consumption.
Charlie Cheng, CEO of TeRAM, said he's grateful for Schechter's conviction in the company's mission and for all the investors supporting this ambitious roadmap. He said developing a new memory technology is a significant undertaking, and that TeRAM is fortunate to have investors backing its vision with both capital and hands-on support.
The funding round was co-led by Primary, B Capital, Hyperion, and SemiAnalysis Capital, joined by Alumni Ventures and Lightscape Partners, and closed on September 10, 2026. TeRAM describes this investor group as bringing deep insight and go-to-market reach across the AI ecosystem, along with broad connections across the capital markets. The funding will let TeRAM validate its technology and scale up its team to build customized 3D SRAM memory chips for its initial data center customers.
TeRAM's founders bring more than 80 years of combined experience developing semiconductor memory technology, and the team has also worked closely on AI compute system-on-chip (SoC) projects with several leading hyperscalers and AI infrastructure companies. This combination of memory and compute expertise is meant to help TeRAM understand the challenges facing future AI platforms and develop custom 3D SRAM tailored to each customer's specific capacity, bandwidth, and power needs.
