XCENA raises $135M betting AI's bottleneck is memory, not compute

XCENA, a South Korean chip startup, just raised $135 million at a $570 million valuation to develop memory solutions for AI. They argue that while most companies focus on improving AI's processing power, the real bottleneck is memory — how quickly and efficiently AI systems can access and store data. In plain English, think of it like a supercar with a tiny fuel tank: no matter how fast the engine is, you’ll always be stuck refueling.

This matters because it could change how AI systems are designed. If memory is the limiting factor, improving it could make AI faster and more efficient for everyday uses like voice assistants, self-driving cars, and even smartphone apps. For example, your phone’s AI features might get a lot faster if they can access data more quickly.

If you're curious about how memory affects AI, try a simple test: ask your smartphone’s AI assistant to do something complex, like summarize a long article. Notice how long it takes? That delay is partly due to memory limitations. Companies like XCENA are working to cut that delay down.