7 hours ago
ARTICLE AD BOX
Intel has travel retired swinging astatine Computex 2025, unveiling a compelling caller scope of graphics processing units (GPUs) and AI accelerators aimed squarely astatine professionals and developers.
The tech elephantine besides marks a important milestone, celebrating 40 years of collaboration and invention wrong Taiwan’s move exertion ecosystem. This day underscores nan enduring value of nan x86 architecture successful driving high-performance computing, particularly arsenic AI applications request ever much processing power.
“For nan past 40 years, nan powerfulness of our business pinch nan Taiwan ecosystem has fueled invention that has changed our world for nan better.
This week, we are renewing our committedness to our partners arsenic we activity to build a caller Intel for nan future. Together, we will create awesome products that delight our customers and capitalize connected nan breathtaking opportunities ahead.”
New Intel Arc Pro B-Series GPUs
Intel is expanding its master graphics lineup pinch nan preamble of nan Intel Arc Pro B60 and Intel Arc Pro B50 GPUs.
These caller cards, built connected nan precocious Xe2 architecture, boast Intel Xe Matrix Extensions (XMX) AI cores and cutting-edge ray tracing units. This operation delivers nan high-performance grunt needed by creators, developers, and engineers tackling demanding tasks.
The Arc Pro B60, equipped pinch 24GB of memory, and nan Arc Pro B50, featuring 16GB, are designed for intensive AI conclusion workloads and master workstation applications. Their AI-ready features and multi-GPU scalability make them a versatile prime for a scope of master users.
These GPUs are specifically optimized for AEC (architecture, engineering, construction) and AI conclusion workstations. They connection nan stableness and capacity demanded by professionals, backed by a wide array of independent package vendor (ISV) certifications and optimized software. Notably, nan Intel Arc Pro B-Series GPUs activity seamlessly pinch some user and master drivers connected Windows.
For Linux users, a containerized package stack simplifies AI deployments, pinch ongoing characteristic upgrades and optimizations planned. By pairing important representation capacity pinch important package compatibility, Intel intends to supply creators and AI developers pinch scalable and cost-effective solutions.
Intel besides previewed Project Battlematrix, a configurable workstation-class level powered by Intel Xeon processors. This level is engineered to streamline nan AI improvement process by supporting up to 8 Intel Arc Pro B60 24GB GPUs.
This configuration allows for nan training of medium-sized (up to 150 cardinal parameters) and highly meticulous AI models, acknowledgment to a monolithic 192GB of video random-access memory.
“The Intel Arc Pro B-Series showcases Intel’s committedness successful GPU exertion and ecosystem partnerships. With Xe2 architecture’s precocious capabilities and a increasing package ecosystem, nan caller Arc Pro GPUs present accessibility and scalability to mini and medium-sized businesses that person been looking for targeted solutions.”
Vivian Lien, vice president and wide head of Client Graphics astatine Intel.Australian professionals tin expect nan Intel Arc Pro B60 GPU to statesman sampling from add-in committee partners for illustration ASRock, Gunnir, Lanner, Maxsun, Onix, Senao, and Sparkle starting successful June 2025. The Intel Arc Pro B50 GPU will go disposable done Intel-authorized resellers starting successful July 2025, pinch section pricing to beryllium announced person to nan merchandise date.
Intel Gaudi 3 AI Accelerators
Intel is besides expanding its AI prowess pinch caller deployment options for its Intel Gaudi 3 AI accelerators.
Intel Gaudi 3 PCIe cards
These cards are designed to facilitate scalable AI inferencing wrong existing information centre server environments. Leveraging AI models for illustration Llama, businesses of each sizes tin use from nan elasticity to tally models ranging from Llama 3.1 8B to full-scale Llama 4 Scout aliases Maverick, acknowledgment to nan adaptable configurations available.
Intel Gaudi 3 PCIe cards are slated for readiness successful nan 2nd half of 2025, pinch Australian readiness and pricing to beryllium confirmed.
Intel Gaudi 3 rack standard system
These reference designs prioritize elasticity and scale, supporting up to 64 accelerators per rack and boasting a monolithic 8.2 terabytes (TB) of high-bandwidth memory. The unfastened and modular creation intends to forestall vendor lock-in, while nan cabled backplane and blind-mate 2D full-rack cabling simplify some installation and maintenance.
Featuring liquid cooling, these systems are engineered to present robust capacity while keeping nan full costs of ownership (TCO) successful check.
The Intel Gaudi rack standard architecture is specifically optimized for moving ample AI models and excels successful real-time inferencing pinch low-latency performance. These configurations reenforce Intel’s dedication to providing open, flexible, and unafraid AI infrastructure, supporting some civilization and Open Compute Project (OCP) designs for unreality work providers (CSPs) successful Australia and globally.
Intel AI Assistant Builder
Following its first unveiling astatine CES 2025, Intel AI Assistant Builder is now publically disposable successful beta connected GitHub.
This lightweight and unfastened package model empowers developers to build and tally civilization AI agents locally connected Intel-based AI PCs. With innovative solutions being showcased astatine Computex by partners for illustration Acer and ASUS, Intel AI Assistant Builder enables developers and partners to quickly create and deploy AI agents tailored to their circumstantial organizational needs and nonstop customer applications. Australian developers tin now entree this instrumentality to research nan possibilities of section AI supplier development.
For much information, caput to Intel.
English (US) · 
Indonesian (ID) · 