MaxLinear, announced it is showcasing Panther V, the latest generation of its storage accelerator platform, at Dell Technologies World 2026. The event takes place May 18–21 at The Venetian, Las Vegas (Booth 317). Panther V addresses data-movement bottlenecks in large-scale AI inference data centers as workloads shift to real-time, production-scale inference. It targets constraints related to the cost, latency, and inefficiency of data movement across storage, memory, and compute.
The platform is optimized for AI inference and Time-to-First-Token (TTFT). It reduces end-to-end latency and improves responsiveness and throughput by tightly coupling CPU, accelerator, and GPU resources. Inline execution of data transformation, compression, encryption, and integrity operations eliminates unnecessary CPU involvement and memory round-trips. This reduces GPU idle time, accelerates time-to-first-token, and frees host CPUs for model execution.
Panther V supports higher concurrency of inference agents, improving utilization and scalability for latency-sensitive applications such as agentic inference.
It is designed for demanding inference scenarios including:
- Low-latency inference for conversational AI and real-time applications
- Retrieval-Augmented Generation (RAG)
- KV-cache-intensive inference
Key capabilities of Panther V include:
- Scalable performance supporting system architectures exceeding 6Tbps, with up to 450Gbps per accelerator
- CPU offload through dedicated hardware engines for single-pass compression, encryption, and checksum processing
- Advanced accelerations: GZIP, Zlib, Deflate, XP10, AES encryption (ECB, CBC, CTR, XTS, GCM), SHA-1/2 hashing and checksums
- Data integrity with real-time end-to-end verification, CRC validation, and NVMe T10 DIF/DIX support
- Software flexibility via SDK with synchronous and asynchronous APIs, kernel and user space support, NUMA-aware queues, and peer-to-peer DMA
- ZFlush for OpenZFS hardware-accelerated implementation
- Industry-standard form factors: PCIe and OCP NIC 3.0
“AI inference is rapidly becoming a real-time, revenue-generating workload, and data movement, not compute, is emerging as the primary system bottleneck,” said Vikas Choudhary, SVP & GM of the Connectivity and Storage Business at MaxLinear. “By accelerating faster node bring-up, growing context sizes, and KV-cache compression, Panther V enables more efficient and low latency inference pipelines along with scalable AI inference economics. We believe that the size of the serviceable market for purpose-built silicon accelerator solutions, such as Panther V, is approximately $5 billion.”
MaxLinear representatives will be available at Dell Technologies World 2026, May 18–21, at Booth 204.
