Intel ships Arc Pro B50 for small-form-factor workstations with local Artificial Intelligence capability

Intel has shipped the Arc Pro B50 workstation GPU to reviewers, a low-profile, small-form-factor card designed for GPU-accelerated workflows including local Artificial Intelligence. The card packs 16 Xe2 cores, 16 GB of GDDR6 memory and draws 70 W, running entirely from the PCIe slot.

Intel has shipped the Arc ´Battlemage´ Pro B50 workstation GPU to reviewers, showcasing the capabilities of the BMG-G21 silicon in a compact, small-form-factor design. The card includes 16 Xe2 cores and 16 GB of GDDR6 memory clocked at 14 Gbps on a 128-bit bus, yielding roughly 224 GB per second of memory bandwidth. The board is a dual-slot, low-profile model that uses only the PCIe slot for power, with a total board power of 70 W and no external power connectors. Signal lanes run at PCIe Gen 5 using eight lanes.

Intel positions the Arc Pro series for a range of GPU-accelerated workflows, explicitly naming Artificial Intelligence, design, engineering, architecture and manufacturing applications. The company pairs the hardware with workstation-focused, validated graphics drivers intended to provide robust, reliable performance in professional environments. Intel asserts the B50 delivers 170 TOPS of INT8 compute and describes it as suitable for local Artificial Intelligence work. The card offers four mini DisplayPort outputs and will be sold either as a standalone package or integrated into OEM systems. Pricing is not provided in the article and is listed as not stated.

The shipment to reviewers is framed as a demonstration of what the BMG-G21 silicon can deliver in a tiny chassis, highlighting trade-offs inherent to low-power, small-form-factor cards. While memory bandwidth, modest core count and 70 W power limit absolute throughput compared with larger workstation cards, Intel argues these characteristics are balanced to keep the 16 Xe2 cores fed and to support local inference and professional workloads in space-constrained systems. The article references Intel´s official performance figures but does not include specific benchmark numbers in the text. Availability timing beyond reviewer shipments is not stated.

63

Impact Score

AMD and Rackspace plan dedicated AI compute rollout

AMD and Rackspace have finalized a phased deployment for dedicated AMD-based compute across Rackspace data centers. The capacity is aimed at regulated enterprise workloads, including clinical AI and large-scale inference.

Lexar tests SSD offloading for local AI models

Lexar is developing an AI-focused SSD approach designed to cut DRAM demand when running large language models on consumer PCs. Internal tests show the company’s storage offloading can load models that traditional local frameworks struggle to run with limited memory.

NVIDIA Blackwell leads MLPerf Training 6.0

NVIDIA’s latest MLPerf Training 6.0 results put Blackwell across every benchmark in the suite, including new MoE workloads. Partner systems from Microsoft Azure and CoreWeave highlighted large-cluster runs on Llama 3.1 405B and DeepSeek-V3 671B.

HPE and NVIDIA expand AI Factory for agentic systems

HPE and NVIDIA are adding agent tooling, confidential computing and updated accelerated systems across the HPE AI Factory portfolio. The expansion targets production deployments that need governance, secure data handling and integrated networking.

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.