Arm

Heading to PyTorch Conference next week? So are we! 🤝 Join us in San Francisco to see how we’re empowering AI developers to build high-performance, energy-efficient GenAI applications across cloud, mobile, and edge using PyTorch and ExecuTorch. Stop by Booth #P1 for hands-on demos, one-on-one workshops, and insightful talks from Arm experts and ambassadors. If you’re in the Bay Area, don’t miss this chance to connect with us in person! #PyTorchCon https://okt.to/cls1Bq

Fresh from OPPO’s Developer Conference and the ColorOS 16 launch, we’re thrilled to share more about our collaboration to integrate SME2 into OPPO's AI framework - unlocking faster performance, greater efficiency, and smoother development. Built into the Arm Lumex compute platform and tested on OPPO AI’s Automatic Speech Recognition, SME2 delivered a 1.2x performance improvement, and a 63% reduction in quantization precision loss, setting new standards for on-device AI. At the event, Steve Raphael, Senior Director in our Client Line of Business, shared how Arm, together with partners like OPPO, is shaping the next generation of intelligent, responsive user experiences through SME2 and KleidiAI. 📸 This collaboration between Arm and OPPO marks another exciting step toward a more connected AI ecosystem - bringing advanced on-device intelligence to billions of users worldwide. 🤝

A new era for AI workstations has arrived. 💻 The Arm-powered NVIDIA DGX Spark is bringing petaflop-scale AI performance to the desktop, enabling developers, researchers, and enterprises to train and deploy models of up to 200 billion parameters, entirely on-device. Built on the NVIDIA GB10 Grace Blackwell Superchip, these new systems from leading OEMs including Acer, ASUS, Dell Technologies, HP, Lenovo, MSI, and more to combine high compute density, unified memory, and native toolchain portability to make large-scale AI faster, more efficient, and more accessible. From enterprise AI development to creative industries, these Arm-based workstations redefine what’s possible from cloud to edge. https://okt.to/V2qdlX

A calm moment before the doors opened at #OCPSummit25. Since then, the Arm booth has been buzzing - packed with great conversations, productive meetings, and insights on building the Converged AI data center 🤝 where performance meets efficiency, and collaboration fuels innovation and scale. Here's to a great final day of the show! 🙌

Designed for AI, Apple’s new M5 chip, built on the Arm architecture, brings major gains in performance and efficiency. Powering the new MacBook Pro, iPad Pro, and Apple Vision Pro, it puts Arm innovation at the center of faster, smarter on-device AI experiences: https://okt.to/nN4r7e

Joining the Open Compute Project Foundation board marks an important step forward in delivering open, efficient solutions for the future of AI infrastructure. At #OCPSummit25, Eddie Ramirez spoke with TechArena.ai about our latest contribution, the Foundation Chiplet System Architecture (FCSA) - a framework designed to accelerate chiplet interoperability and unlock new innovation across the data center ecosystem. Together with Arm Total Design, FCSA empowers companies of all sizes to innovate faster, scale efficiently, and bring inference closer to the edge. As Eddie says, “Efficiency isn’t just an advantage anymore; it’s a survival requirement.” Full piece here: https://okt.to/IKOHDG

AI at the edge is evolving rapidly. In the latest episode podcast episode of Arm Viewpoints, Arm’s Paul Williamson, SVP and GM of IoT, sits down with Chris Rommel from VDC Research to explore how AI is reshaping embedded and IoT development, and what it means for engineering organizations worldwide. From: ▪️How developers are adapting to the complexity of Edge AI workloads ▪️The critical role of software tools and ecosystem collaboration ▪️Why flexibility, scalability, and trust are key to future-ready designs The two are diving into it all. This conversation circles back to the findings from VDC’s latest research report, and offers a firsthand look at how Arm is empowering developers and partners to innovate with AI at the edge. 💪 https://okt.to/cqs2Xl

As vehicles become increasingly AI-defined, the automotive industry needs new levels of performance, efficiency, and design flexibility. 🚘 This week, we took a major step forward in shaping that future by contributing the Foundation Chiplet System Architecture (FCSA) to Open Compute Project Foundation (OCP). FCSA creates a common framework for chiplet interoperability, helping OEMs, tier-1 suppliers, and vendors alike accelerate innovation, reuse IP, and scale designs across multiple vehicle platforms. By advancing FCSA through OCP, we’re helping to build an open, collaborative chiplet ecosystem that can support the growing complexity of automotive compute. This is just the beginning of a future defined by centralized compute and AI-driven vehicles. https://okt.to/EV3wqB

Here’s to great conversations at #OCPSummit25! 💡 Eddie Ramirez was joined by Santhosh Srinath (Meta), Jinwook Oh (Rebellions), and Daniel Ping (Novatek) to discuss the technology and collaboration shaping the next generation of composable, AI-ready infrastructure. Together, they explored how modular, chiplet-based architectures are coming to life on Arm - and what that means for the next wave of AI and general purpose compute. From AMBA CHI C2C open standards to advances in packaging, system integration, and lifecycle analytics, the conversation highlighted innovation and collaboration across the Arm ecosystem - and the incredible momentum behind it.

#OCPSummit25 wouldn’t be complete without our hardware and DPU wall, showcasing the incredible innovation of our ecosystem partners. We've got a seriously impressive line up, featuring: 🔹 NVIDIA Grace Blackwell GB300 🔹 NeuReality NR1 🔹 Novatek’s Arm Neoverse CSS N2-based SoC 🔹 …and many more from Marvell, xSight, and others What catches your eye first?