Nvidia debuts deskside supercomputers for frontier Artificial Intelligence models

Nvidia introduced its DGX Spark and DGX Station deskside supercomputers at CES, targeting developers working with open-source and frontier Artificial Intelligence models that need to run locally before scaling to the cloud.

Nvidia is positioning its new DGX Spark and DGX Station systems as deskside supercomputers designed for developers working with open-source and frontier Artificial Intelligence models. Announced at the CES trade show, the machines are aimed at organizations and teams that want high performance model development capabilities in a local environment instead of relying solely on remote data centers.

The company says the DGX Spark and DGX Station let developers run the latest open and frontier Artificial Intelligence models directly on a deskside system, with support ranging from 100-billion-parameter models on DGX Spark to 1-trillion-parameter models on DGX Station. By offering this level of capacity in a form factor that can sit near a developer’s workspace, Nvidia is trying to shorten iteration cycles for training, fine-tuning, and experimentation while maintaining access to very large model sizes.

Both systems are powered by the Nvidia Grace Blackwell architecture, which is described as combining large unified memory with petaflop-level Artificial Intelligence performance. Nvidia presents this combination as giving developers new capabilities to develop applications locally while retaining a straightforward path to scale workloads to the cloud. The focus is on enabling a workflow where teams can prototype, refine, and validate models on DGX Spark or DGX Station, then extend or deploy those models across larger infrastructure as demand grows.

68

Impact Score

Huawei chip design raises pressure on Nvidia, AMD, and Intel

Huawei has outlined a new chip design framework that it says can improve efficiency and reduce dependence on leading-edge manufacturing tools. The move adds pressure on US chipmakers as China builds a domestic Artificial Intelligence semiconductor ecosystem under export restrictions.

UK and EU seek simpler medical device rules

The UK and EU are advancing medical device regulatory changes aimed at improving predictability, reducing bottlenecks and supporting market access. Manufacturers of Artificial Intelligence-enabled devices in Europe will still need to navigate overlapping rules even as compliance timelines are extended.

LLMSurgeon targets foundation model data auditing

LLMSurgeon introduces a way to infer the domain mix of large language model pretraining data using only generated text. The framework is designed to improve transparency around foundation models whose training corpora remain largely undisclosed.

Databricks model units target lower inference costs

Databricks is positioning model units as a new way to manage large language model inference, aiming to cut GPU spending while improving reliability under enterprise-scale demand. The approach reflects growing pressure on platforms to balance cost, latency, and resilience as agentic Artificial Intelligence workloads expand.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.