Intel has launched its Arc Pro B60 and B50 GPUs on May 19, 2025, ahead of Computex, targeting workstations and AI inference with increased memory and software support.
At Computex in Taipei (May 20-23), Intel will celebrate 40 years of local partnerships and showcase its expanded GPU offerings, AI accelerator advancements, and wider availability of its AI assistant.
Intel Arc Pro B60 and B50 GPUs
Intel has introduced the new Intel Arc Pro B60 and B50 GPUs, built on the Xe2 architecture, featuring Intel Xe Matrix Extensions (XMX) AI cores and advanced ray tracing units. These GPUs are tailored for high-performance AI workloads and workstation applications, offering 24GB and 16GB memory, respectively. Additionally, Intel unveiled a configurable workstation-class platform (Project Battlematrix) that integrates up to eight Intel Arc Pro B60 GPUs to handle AI models with up to 192GB of VRAM.
The Intel Arc Pro B-Series GPUs provide a scalable and cost-effective solution for creators, developers, and engineers by supporting demanding AI inference workloads and delivering optimized performance for AEC (Architecture, Engineering, and Construction) and other workstation tasks. The GPUs ensure software stability through ISV certifications and compatibility with both Windows and Linux environments.
Intel’s strategy focuses on expanding its GPU lineup for professional markets and integrating AI-ready features into its platforms. The company is also enhancing its software ecosystem and establishing partnerships with key board partners such as ASRock, Gunnir, Lanner, Maxsun, Onix, Senao, and Sparkle to drive adoption. The Arc Pro B60 and B50 GPUs will be available starting in June and July 2025, respectively.
Intel Gaudi 3 AI accelerators
Intel has expanded its AI strategy with the introduction of Intel Gaudi 3 AI accelerators, available in PCIe card format and rack scale system configurations. The PCIe cards enable scalable AI inferencing within existing data center infrastructure, supporting AI models from Llama 3.1 8B to Llama 4 Scout and Maverick. The rack scale system offers up to 64 accelerators per rack and 8.2 TB of high-bandwidth memory, featuring an open, modular design and liquid cooling for enhanced performance and cost efficiency.
Intel Gaudi 3 provides scalable AI deployment options for diverse customer needs, from small businesses to large enterprises. The PCIe cards offer flexibility and integration with existing systems, while the rack scale system excels in running large AI models with low-latency performance. Both configurations support open standards, reducing vendor lock-in and lowering the total cost of ownership (TCO).
Intel’s strategy emphasizes open, flexible, and secure AI infrastructure, targeting cloud service providers and enterprise customers. By offering modular, scalable designs that align with Open Compute Project (OCP) standards, Intel is positioning Gaudi 3 as a versatile solution for AI deployments in data centers, reinforcing its commitment to accessible AI acceleration.
Intel’s AI Assistant Builder
Intel has launched the AI Assistant Builder, a lightweight, open-source framework designed to build and run custom AI agents locally on Intel-based AI PCs. Following its debut at CES 2025, the framework is now available in public beta on GitHub. It allows developers and partners to rapidly create and deploy AI agents, with early examples showcased by Acer and ASUS at Computex 2025.
The Intel AI Assistant Builder provides a cost-effective and accessible tool for organizations to develop AI agents tailored to specific business needs. By enabling local deployment, it ensures data privacy and reduces latency. The open-source framework fosters flexibility and rapid prototyping, making it ideal for small businesses and large enterprises alike.
Intel’s strategy focuses on empowering developers and partners to build AI capabilities directly into Intel-based systems, expanding the AI ecosystem across consumer and enterprise markets. By leveraging open-source accessibility and strategic partnerships with key OEMs, Intel aims to accelerate AI adoption and establish its hardware and software solutions as foundational components for AI development.
Baburajan Kizhakedath
