The big picture: XCENA, a chip startup with offices in South Korea and the US, has raised $135 million in a Series B at a $570 million valuation. The round was led by Seoul-based firms Atinum and IMM Investment.
Why it matters:
- Memory bottleneck: XCENA argues that AI inference is increasingly constrained by memory scaling, not just raw compute power, as data volumes grow.
- Efficiency gains: Its MX1 chip processes AI data directly within memory, bypassing expensive CPUs and GPUs for preprocessing, KV cache management, and data orchestration.
- Infrastructure reduction: The company claims this in-memory processing approach could reduce server requirements tenfold, significantly cutting AI infrastructure costs.
How it works:
- Integrated processing: The MX1 chip is designed to connect to processors via CXL, handling critical AI data tasks directly inside the memory module.
- Samsung partnership: Mass production of the MX1 chip is scheduled via Samsung’s foundry by the end of 2026.
- Market entry: XCENA expects to begin generating revenue from its in-memory AI chips starting in 2027.
The catch: The AI chip market is highly competitive, with established players and numerous startups vying for market share. XCENA’s success will depend on its ability to demonstrate significant performance and cost advantages, secure design wins with major customers, and navigate the complexities of mass production and supply chain management.
Key Facts
- Company: XCENA
- Amount: $135M
- Round: Series B
- Investor: Atinum and IMM Investment (co-lead)
- Founders: Jin Kim, Dohun Kim, Harry Juhyun Kim
- Valuation: $570,000,000
- Announced: 2026-06-02
- Sector: AI Hardware
- Headquarters: Seoul, South Korea
