By: Jay Yeon, Senior Marketing Manager
CXL Consortium member XCENA recently participated in a Q&A session to discuss CXL use cases and the benefits for hyperscale data centers, database companies, and High-Performance Computing industries.
Can you please tell the CXL community about XCENA ?
XCENA is a South Korea-based fabless startup tackling the memory wall limiting modern data-driven systems. We develop computational memory and memory pooling products based on our HW-SW Co-optimized Architecture supporting CXL® standards. Our products target data-heavy domains like AI, big data, vector databases, and data analytics acceleration.
Our CXL 3.x-enabled products bring compute and data closer together by allowing more data in memory, eliminating bottlenecks and improving performance.
Computational Memory and Memory Pooling with SSD Backend
What prompted XCENA to join the CXL Consortium?
Joining the CXL Consortium was a natural step for XCENA as our technology roadmap is closely aligned with the fundamental goals of CXL. We recognized that advancing open, coherent memory interconnects requires collective industry alignment. By engaging directly in the consortium, we ensure our memory-centric innovations influence evolving CXL standards and stay aligned with ecosystem requirements.
What do you feel are the main advantages of becoming a CXL Consortium member? And can you share how XCENA is participating?
CXL Consortium membership allows us the opportunity to engage directly with a global network of industry leaders spanning semiconductors, hyperscalers, server OEMs, accelerator developers, and broader ecosystem players. XCENA’s membership enables us to:
- Align our development roadmap with emerging use cases and best practices across the CXL ecosystem.
- Accelerate product validation and ensure interoperability through collaboration with processor vendors, platform developers, and system integrators.
- Engage with potential partners and customers to explore real-world CXL use cases and strengthen our presence in the community.
XCENA actively contributes to the ongoing discussions around memory architecture and we are showcasing our CXL computational memory at industry events such as VeloxCon (Link: https://www.youtube.com/watch?v=hMn3vs0c6q8), and others. We also plan to share empirical insights from our prototypes in the near future.
Are there any use cases you feel would see particular benefits from CXL technology?
We see several high-impact use cases for CXL-enabled solutions:
- Vector database acceleration – These workloads require large, low-latency memory pools to support fast similarity search across high-dimensional data. Our CXL-enabled solutions are purpose-built for vector DB acceleration, supporting memory scaling and accelerating data processing.
- AI training and inference – These workloads require scalable memory to support growing model sizes and high-throughput compute. CXL-enabled solutions serve as memory expanders, alleviating bottlenecks and enabling efficient resource scaling across the AI pipeline.
- Near-data processing – CXL enables intelligent memory nodes that can offload computation closer to data with cache coherency, improving performance and power efficiency while reducing total cost of ownership. Combining CXL-based memory expansion with near-data processing offers a powerful solution to the memory wall challenge in modern computing architectures.
Which market segments do you believe will benefit most from CXL deployment?
We see the following market segments as poised to gain the most from CXL:
- Hyperscale data centers – CXL enables more scalable and cost-efficient infrastructure through memory disaggregation and resource pooling.
- Database companies – Expanding and sharing memory across CPUs and accelerators helps overcome architectural constraints in large-scale data processing.
- High-performance computing – CXL-enabled memory expansion with near-data processing can reduce bandwidth consumption and improve throughput for memory-bound workloads.
How is CXL impacting your business?
CXL plays a foundational role in XCENA’s product and business strategy. Our R&D—including CXL 3.x silicon and SDKs—is directly aligned with Consortium specifications. It also drives co-design collaboration with CPU vendors, system architects, and hyperscalers to ensure seamless integration.
As the ecosystem matures, we remain focused on delivering data-centric solutions that improve performance and reduce total cost of ownership.
Where can we learn more about XCENA?
Learn more about XCENA by visiting our website (https://xcena.com/) and YouTube channel (@XCENAOnYoutube). Follow us on LinkedIn for updates.