Last Updated: 29 May 2026
XCENA, a South Korean semiconductor company building memory-centric computing solutions for AI infrastructure, has closed a $135 million Series B round. The company was founded by veterans of Samsung and SK Hynix and is led by CEO Jin Kim, who previously held a Corporate VP role at SK Hynix, where he was one of the youngest executives to reach that position.
The fundraise comes as memory is increasingly becoming the limiting factor in AI system performance. Prices for high-bandwidth memory surged 246% year-over-year in 2025, and suppliers are effectively sold out through 2026. The total semiconductor market for data centers is projected to grow from $209 billion in 2024 to $492 billion by 2030, with memory architecture at the center of that shift.
XCENA has now raised $185 million in total and carries a valuation of $570 million. The Series B was co-led by Atinum Investment and IMM Investment, with a mix of new and returning backers from across Asia's venture and financial sectors.
For most of the past decade, AI improvements were driven by faster processors. More compute power meant better results. That logic still holds, but a separate problem has grown alongside it: getting data from memory to the processor fast enough to avoid delays. When the processor sits idle waiting for data to arrive, performance suffers regardless of how powerful the chip is. This is what engineers call the memory wall.
The gap between processing speed and memory speed has become one of the main engineering challenges in modern AI infrastructure. CXL, or Compute Express Link, is an open industry standard gaining adoption as a way to address this. It lets memory be pooled and shared more flexibly across processors, reducing the distance data has to travel and cutting down on wasted time and energy.
XCENA builds its products on the CXL 3.x standard. The company is focused on what it calls computational memory, an approach where some processing happens directly inside the memory unit rather than waiting for data to travel to a separate chip.
"XCENA is redefining how computational memory is applied in real-world systems. Their MX1 product is already showing how customers can simplify complex infrastructure, accelerate deployments, and eliminate inefficiencies that have traditionally slowed down advanced computing workflows. We're excited to support XCENA as it scales these capabilities globally."
Sangmin Lim, Investment Director, Atinum Investment
The funding will go toward three areas. XCENA will expand customer deployments and deepen validation of the MX1 with enterprise customers and ecosystem partners. It will also build out its sales and partnership infrastructure to reach hyperscalers, telecoms, and research institutions. In parallel, development of its next generation of computational memory products will continue.
The company is growing its presence in Northern California to work more closely with large technology companies and AI infrastructure buyers in the US market. XCENA also says it is in active discussions with international institutional investors as part of its broader expansion plans.
The company was originally founded under the name MetisX and later rebranded. It is headquartered in Seongnam, South Korea, and has a US office in Sunnyvale, California. Its founding team came from SK Hynix and Samsung, and the broader leadership group brings decades of experience in semiconductor hardware.
The MX1 is XCENA's main product. It combines a large pool of DDR5 memory with processing cores placed physically close to that memory. The idea is straightforward: rather than moving large amounts of data to a distant processor, some calculations happen right where the data already is. That reduces both the time it takes and the energy consumed, which matters most for demanding workloads like AI inference, large-scale database queries, and high-performance computing.
"AI workloads are exposing the fundamental limitations of traditional computing architectures as larger models, expanding context windows, and increasingly data-intensive inference workloads drive unprecedented memory demands. With strong backing from leading global investors, we are accelerating delivery of MX1 into emerging AI infrastructure ecosystems and advancing the next wave of memory-centric computing systems."
Jin Kim, CEO and cofounder of XCENA
XCENA also provides a full-stack software development kit, or SDK, that lets engineers integrate the MX1 into existing systems without rebuilding their architecture from scratch. The company says this is intended to make adoption faster and less disruptive for customers.
The Series B drew participation from a large group of Korean financial institutions and venture firms. Returning investors include SBI Investment, Mirae Asset Capital, Mirae Asset Venture Investments, STIC Ventures, Wonik Investment Partners, SV Investment, and LB Investment.
New participants in the round include Corstone Asia, Kiwoom Investment, DSC Investment, Shinhan Venture Investment, Korea Development Bank, KDB Capital, Premier Partners, Kolon Investment, Company K Partners, K2 Investment Partners, Partners Investment, and Kyobo Securities and Kyobo Life. The presence of state-backed institutions such as Korea Development Bank alongside private firms reflects the breadth of the investor lineup.
XCENA's total funding now stands at $185 million.
