Dell and Nvidia Introduced Hybrid AI Stack

The new infrastructure aims to reduce cloud costs by enabling local agentic AI processing for enterprises.

Updated on Sept. 23, 2026 in Artificial Intelligence

Isometric editorial illustration showing a high-performance workstation and server unit connected by geometric data conduits, representing a hybrid AI infrastructure. — Dell and Nvidia have introduced a new hybrid AI infrastructure stack designed to help enterprises reduce costs by processing agentic AI tasks locally. AI Illustration. Upload story photo >

Live Poll

Do you trust that businesses can keep the costs of adopting AI technologies under control?

Dell and Nvidia have launched a hybrid AI stack designed to allow businesses to run open models locally. This architecture leverages the Dell Pro Max with GB10 workstation to manage agentic tasks more efficiently than cloud-only solutions.

Why it matters

Rising token consumption in public-cloud APIs has led to unpredictable enterprise operating costs. This hybrid approach provides a secure and governable infrastructure to help companies scale their agentic AI systems economically.

The Dell Pro Max with GB10 workstation features the NVIDIA GB10 Grace Blackwell Superchip and 128GB of unified memory. The NemoClaw reference stack uses OpenShell technology to isolate and govern agent activity.

The players

Dell Technologies

This multinational technology company provides hardware and software solutions and is a major manufacturer of enterprise server and workstation infrastructure.

Nvidia

A global leader in accelerated computing, this company designs the graphics processing units and AI chips that power modern generative AI infrastructure.

The details

Dell tested agentic AI tasks using two different methods, finding that utilizing a coordination agent significantly reduced token usage compared to large context windows. The hybrid architecture allows for local model execution while maintaining cloud access for frontier models.

Two years ago, hyperscaler AI capital expenditure was estimated at USD 250 billion.
In 2026, hyperscaler AI capital expenditure surpassed USD 800 billion.

The Tech Race

As hyperscaler AI capital expenditure has soared to USD 800 billion, firms are moving toward hybrid architectures to maintain control over rising operational costs. This shift represents a transition from centralized cloud reliance to a model that balances local edge computing with cloud-based frontier intelligence.

Enterprises can expect more predictable budgeting for AI projects as infrastructure shifts toward local agentic processing. IT departments will need to manage new hardware, such as the GB10 workstations, to support these governable AI workflows.

The takeaway

Businesses looking to optimize AI spending should evaluate whether their current workloads are better suited for local agentic execution rather than cloud-only APIs. This hybrid strategy allows for greater security and control over sensitive enterprise data as reasoning usage continues to rise.