Xcena raises $135M betting AI’s bottleneck is memory, not compute
Original: This chip startup just raised $135M on a bet that AI’s biggest bottleneck isn’t compute — it’s memory
Xcena raised $135 million to build memory-centric chips aimed at cutting AI inference infrastructure costs.
South Korean chip startup Xcena raised a $135 million Series B at a $570 million valuation, bringing total funding to $185 million. The company argues AI inference is increasingly constrained by memory movement, not just GPU compute. Its prototype MX1 chip uses CXL to process data closer to DRAM, with Samsung foundry mass production planned by late 2026 and revenue targeted for 2027.
TechCrunch reports that Korean chip startup Xcena is betting that the real bottleneck in AI infrastructure is shifting from pure compute to memory. Every time you use a generative AI service like ChatGPT, data moves back and forth among memory, the CPU, and the GPU; generating each word of the model can repeat this data-shuttling process. Xcena's argument is that this structural round-tripping of data is not only expensive but also power-hungry, and in large-scale AI inference scenarios in particular it magnifies into a cloud-cost problem.
Free shows the 3-line summary; Pro unlocks the full deep summary (~300 words) so you never have to click through.
See Pro plans →Want the original English / full article?
Read on TechCrunch AI →Summaries are AI-generated; the original article is authoritative.