Chip startup XCENA raises $135M betting memory is AI's bottleneck
Original: This chip startup just raised $135M on a bet that AI’s biggest bottleneck isn’t compute — it’s memory
Why This Matters
Addresses critical AI infrastructure inefficiency by reducing memory-compute bottlenecks
XCENA, a South Korean-U.S. chip startup, raised $135 million Series B at $570 million valuation. The company designs chips that place compute capabilities closer to DRAM memory to reduce data transfer inefficiencies in AI systems.
XCENA, founded in 2022 by Samsung and SK Hynix veterans, addresses AI infrastructure inefficiencies where data must travel between memory, CPUs, and GPUs for every AI request. Their MX1 chip connects to CPUs through CXL technology, processing data within memory modules rather than requiring costly round trips. CEO Jin Kim claims systems that previously required 10 servers could run on one. The startup has raised $185 million total, betting on memory-centric AI architectures as major memory companies Samsung, SK Hynix, and Micron each crossed trillion-dollar valuations. The chip handles preprocessing, KV cache management, and data caching directly within memory modules.