Dell Technologies expanded its Dell AI Factory portfolio at NVIDIA GTC Berlin, introducing a 48U PowerEdge M9882 rack housing 256 NVIDIA Vera CPUs for agent workload execution, the PowerEdge XE9810 rack-scale inference system built around Groq 3 LPUs, an air-cooled two-socket Vera server, a 64GB Dell Pro Precision with GB10, plus validated software and deskside deployment templates. The software is available immediately, while new hardware will ship from this month through early 2027. Dell notes the platform now serves over 6,500 customers, up from 5,000 last quarter.

Dell frames this expansion around total cost of ownership, referencing a September 2026 Signal65 and Futurum Group analysis claiming its AI Factory can be up to 90% more cost-effective than public cloud over two years for deskside and datacenter knowledge, sales and software development workloads. It also cites Goldman Sachs and Gartner forecasts that agentic AI will multiply token consumption 24 times while inference costs drop more than 90% by 2030. These are third-party and vendor figures and were not independently audited.

“Enterprises used to think about AI infrastructure as an IT budget question, but that’s changed. AI infrastructure is no longer a cost center. It’s a value creation center,” said Michael Dell, chairman and chief executive officer of Dell Technologies.

PowerEdge Expands CPU Execution and LPU Inference

The PowerEdge M9882 loads 256 NVIDIA Vera CPUs into one 48U rack, built for the CPU workloads of agentic AI: orchestration, tool calling, code execution and evaluation, leaving model inference to GPU clusters. Each Vera CPU has 88 Olympus cores, delivering 22,528 cores per full rack. NVIDIA’s reference design calls for 400TB LPDDR5X, 300TB/s aggregate memory bandwidth and 64 BlueField-4 DPUs per rack; Dell has not released final memory or networking specs for the M9882.

Dell’s rationale for this CPU-dedicated rack centers on power and cooling. Isolating agent execution from GPU clusters lets operators right-size each pool, improving power and cooling efficiency versus mixed, over-provisioned servers. The M9882 is the second standalone agent CPU rack unveiled this month and is scheduled for Q1 2027.

The PowerEdge R8822 brings Vera CPUs into standard datacenters as an air-cooled 2U two-socket server for conventional 19-inch racks. Target use cases include agent orchestration, sandboxing, reinforcement learning and data analytics, and the chassis supports up to four double-wide GPUs. It ships this month, making it the first of these new PowerEdge platforms to reach customers.

The PowerEdge XE9810 is Dell’s first Groq LPU rack-scale inference system, with 16 direct-liquid-cooled compute trays powered by Groq 3 LPUs. Positioned as a disaggregated complement to GPU clusters for production agentic AI, its core design goal is consistent time-to-first-token at large scale. Dell has not published per-tray LPU counts, throughput numbers or full configuration details, with availability set for Q1 2027.

Dell’s Modular Data Center with NVIDIA DSX POD is a containerized reference design, pre-integrated and factory-validated with compute, storage, networking, rack power and liquid cooling. Sites only need to connect the containers to pre-prepared power feeds, cutting deployment on a ready site from six months down to 14 days; capacity scales by adding more containers. It is targeted for early 2027.

For telecom carriers, Dell PowerEdge servers hosting NVIDIA Aerial RAN Computer platforms run RAN and AI workloads on shared hardware, delivering lower TCO compared with separate siloed systems. Dell says this ARC support on PowerEdge is coming soon.

Deskside Systems Add Memory and Deployment Templates

The Dell Pro Precision with GB10 gets a 64GB unified-memory configuration starting October 23, matching the release of the $4,999 NVIDIA DGX Spark 64GB unit. It targets individual developers and small teams running open-weight models locally, and Dell has not yet released pricing for this 64GB SKU.

Clustering supports two 64GB nodes or four 128GB nodes. Users can begin with a single unit and scale out to an eight-node GB10 cluster for larger models and higher concurrent user counts. Internal testing across eight GB10 nodes, two of them Dell units, achieved 24.9GiB/s read throughput from one Solidigm QLC storage array.

A Windows-based Dell XPS 16 Creator Edition with RTX Spark N1X and GB300 Pro is also on the roadmap, offering up to 748GB coherent memory and 20 petaflops of FP4 compute, bringing a deskside Grace Blackwell Ultra system into familiar Windows enterprise environments. No firm release date is available yet.

Dell’s Deskside Agentic AI reference templates are live on GitHub. Five launch blueprints cover agentic coding, agentic VSS, NVIDIA NemoClaw, the Hermes Agent on NemoClaw and a higher-education AI lab stack. Each defines the required hardware, models and software stack for its use case, scaling from one developer to 1,000-plus nodes and supporting up to 96 developers on a single platform.

Software Validation and Governance Across the Stack

Reflection’s Beam, an Apache 2.0 open model built for software engineering and long-horizon agent tasks, is validated and available on the platform today, alongside a knowledge assistant and a Mistral-based agentic AI stack. Dynatrace AI observability and Fogsphere computer vision will arrive in Q4 2026, while SUSE Linux joins the supported OS roster in early Q1 2027.

Dell’s agent security framework has three stacked layers: Dell hardware-rooted infrastructure, Red Hat AI Factory software for platform orchestration, and NVIDIA’s OpenShell runtime architecture for governance across runtime, network, data and infrastructure. Dell Managed Services for AI launches in Q1 2027, adding resource tuning, access controls, consumption tracking and cost management. This lets development teams self-provision AI resources while IT retains oversight and financial guardrails.

Beijing Qianxing Jietong Technology Co., Ltd.
Sandy Yang/Global Strategy Director
WhatsApp / WeChat: +86 13426366826
Email: yangyd@qianxingdata.com
Website: www.qianxingdata.com/www.storagesserver.com
Business Focus:
ICT Product Distribution/System Integration & Services/Infrastructure Solutions
With 20+ years of IT distribution experience, we partner with leading global brands to deliver reliable products and professional services.
“Using Technology to Build an Intelligent World”Your Trusted ICT Product Service Provider!