At SC25 Dell announced Project Lightning to parallelize PowerScale OneFS using pNFS with a separate metadata server and Flex Files for parallel I/O, plus NIXL (Nvidia Inference Transfer Library) support to enable scalable KV cache offload for AI inference; ObjectScale gained S3 tables and vector search APIs to speed analytics and RAG workloads. Dell highlighted a 1-second time-to-first-token at a 131,000-token context window (claimed as 19x faster than standard vLLM), introduced software-only PowerScale subscriptions on qualified PowerEdge servers, and said ObjectScale and PowerScale with Nvidia Dynamo are available now, with software-defined PowerScale coming 1H26 and parallel NFS support in 2026. The announcements deepen Dell’s and Nvidia’s enterprise AI stack integration, aimed at accelerating pilot-to-production adoption, growing recurring software revenue, and intensifying competition in AI-optimized storage for large-model and retrieval workloads.
Dell's SC25 announcements center on Project Lightning, an ~18-month initiative that adds parallel I/O to PowerScale OneFS via pNFS, a separate metadata server (MDS) and Flex Files to distribute data across nodes; Dell also added NIXL (Nvidia Inference Transfer Library) support to enable KV cache offload for inference workloads, claiming a 1-second time-to-first-token (TTFT) at a 131,000-token context window — 19x faster than standard vLLM. PowerScale is being offered in software-only subscription form on qualified PowerEdge servers (e.g., R7725xd), and Dell says ObjectScale and PowerScale with Nvidia Dynamo are globally available now. ObjectScale received AI-optimised search features (S3 tables and vector search APIs) to accelerate analytics, inferencing and RAG directly on object storage; software-defined PowerScale is slated for 1H26 and parallel NFS support for 2026, which staggers full feature availability. Jeff Clarke framed these advances as enabling faster pilot-to-production transitions, implying potential uplift to recurring software revenue and stronger server attach for Dell. Ecosystem signals show broad industry alignment (multiple vendors supporting Dynamo), which both validates the approach and increases competition. Key risks are adoption friction and the multi-stage availability timeline; sentiment metrics are mildly positive overall (0.35) with stronger per-ticker signals for DELL (0.7) and NVDA (0.5), suggesting near-term positive bias for Dell/Nvidia stack exposure if benchmarks and enterprise deployments materialize.
AI-powered research, real-time alerts, and portfolio analytics for institutional investors.
Request a DemoOverall Sentiment
mildly positive
Sentiment Score
0.35
Ticker Sentiment
