AMD and OpenAI Partner for AI Infrastructure

My latest PC build was the muse for a new systems-level analysis I've been wanting to write. As I was mapping out parts for a machine built for gaming, multimedia, and AI, I was struck by how completely the balance of power inside a computer has shifted. The tech in our high-end desktops is a direct extension of a fundamental shift happening in the data center, and it’s re-writing the rules for everyone. The old hierarchy that defined computing for decades, the CPU as king, is over. But the new story isn't a simple "GPU-first" inversion; it's a much more profound transition to a "coherency-first" model. The CPU's critical new role is as a sophisticated data-plane orchestrator, a master conductor feeding specialized hardware and running system-level logic while GPUs handle the heavy computational workloads. The battlefield has moved from the chip to the interconnects. True performance is no longer defined by one component, but by the high-bandwidth "fabric" that binds the CPU, GPU, and accelerators into a single, coherent system. This is why Apple's M-series is so effective and why NVIDIA's Grace Hopper and AMD's MI300A are built as integrated "Superchips." The system itself has become the product. This is a major platform shift. In the full article, I explore what this really means for the industry. I discuss the real competitive moat, which isn't just silicon but NVIDIA's 15-year-old proprietary software ecosystem. I also look at AMD's position, which "didn't miss the moment" thanks to its architectural marvel, the Instinct MI300A. Finally, I cover Intel's $100B foundry strategy as an aggressive pivot to become a systems foundry for the AI era. This deep integration, enabled by catalysts like UCIe, is rewriting the very definition of a "system," and every major player has their own pen to the page. Check out the full article on my Substack: https://lnkd.in/et3W_rKG #aiinfrastructure #computefabric #nvidia #intel #amd #chiplets #ucie #semiconductors #techstrategy #architecture #datacentriccompute #systemdesign #acceleratedcomputing #hardwarerevolution #platformshift

Oracle to deploy 50,000 AMD chips for AI workloads : https://lnkd.in/eyTGHvpK AMD’s Instinct MI450 chips will be released next year. Oracle will deploy approximately 50,000 of them starting in the third quarter of 2026. Oracle is thus expanding its own portfolio beyond existing investments in NVIDIA hardware. However, those who opt for an AI stack with open standards will be able to benefit from the competition between AMD and Nvidia. Not only that, but the MI450 is expected to be released earlier than Nvidia’s Vera Rubin GPUs. Existing Nvidia customers will not be able to simply abandon the ecosystem surrounding CUDA, the programming language and development platform for Nvidia equipment. AMD offers ROCm as an open-source alternative and has matured the technology, but CUDA-driven lock-in keeps Nvidia in the saddle. AI developers often simply have more knowledge of CUDA than other options. #semiconductor #manufacturing #technology #innovation #chips  #semiconductormanufacturing #advancedtechnology #engineering #lithography #nanometer #research #development #AI #EUV #DUV #mobileprocessors #VLSI

NVIDIA has officially partnered with Samsung Foundry to support its growing demand for custom CPUs and XPUs, integrating the Korean giant into its NVLink Fusion ecosystem. Announced at the OCP Global Summit, this collaboration enables Samsung to provide end-to-end design and manufacturing services for custom silicon, including design, verification, and tape-out. While not directly producing NVIDIA’s own chips, Samsung’s inclusion signals a breakthrough in AI infrastructure development and opens the door for other firms like OpenAI to leverage NVIDIA’s ecosystem through Samsung’s advanced 2nm SF2 process. #NVIDIA #ChipMaker #TechGiants #SamsungFoundry #CustomSilicon #AIInfrastructure #NVLinkFusion #Chips #Semiconductors #OCP2025 #2nm #Technology #TechPartnership #ChipDesign #SemiconductorIndustry #TechnologyNews https://lnkd.in/d9nT8GwN

𝐓𝐡𝐞 𝐃𝐫𝐞𝐚𝐦 𝐨𝐟 𝐌𝐨𝐫𝐩𝐡𝐂𝐨𝐫𝐞: 𝐀 𝐓𝐡𝐨𝐮𝐠𝐡𝐭 𝐄𝐱𝐩𝐞𝐫𝐢𝐦𝐞𝐧𝐭 What if a processor didn't have to choose between being a CPU or a GPU? What if it could become both, on the fly? This is the vision of Software-Defined Silicon, which I call the MorphCore. It's a speculative look at a future of adaptive computing. The Core Idea: A Fluid, AI-Orchestrated Fabric Imagine a chip made not of fixed cores, but a sea of tiny, generic compute tiles, Lego bricks of logic. An onboard AI "Conductor" would do more than schedule tasks. It would physically architect the hardware in real-time. - Need a CPU? The AI links tiles into a deep, sequential pipeline. - Need a GPU? It re-wires the same tiles into a broad, parallel grid. This is silicon that reconfigures its identity to match the software's soul. The Potential Outcome: Your device wouldn't have a static "8-core" specification. It would have a pool of computational resources, dynamically sculpted by AI. - Booting your OS → configured for rapid serial tasks. - Launching a game → shifts to a parallelized grid. - Heavy multitasking → a nanosecond ballet of reconfiguration. The chip becomes an intelligent, shape-shifting fabric. The Reality Check: We are nowhere near this today. The challenges are monumental: - The physics of low-latency, low-power reconfiguration are daunting. - Managing cache coherence in a fluid topology is a known hard problem. - The economic viability of a generalized silicon fabric is questionable. Current technologies like AMD's APUs, Intel's tile-based designs, and FPGAs are steps in this direction, but operate at a much less dynamic level. Yet, I find this concept compelling. Maybe the next revolution isn't more cores, but cores that can become anything we need. #Innovation #AI #Hardware #Semiconductors #FutureOfCompute #VLSI #ComputerArchitecture #ArtificialIntelligence #Tech #NextGenComputing #CPU #GPU AMD NVIDIA Intel Corporation

🚀 How NVIDIA Became the Leader in Digital Infrastructure for Data Centers In recent years, NVIDIA has evolved from a leading provider of graphics processing technology into a powerhouse in digital infrastructure, especially in the data center industry. The secret? 🤔 The ability to interconnect thousands of GPUs across multiple racks, operating as a unified mega computer. This scalable architecture accelerates AI applications, scientific simulations, and Large Language Models (LLMs) at unprecedented speeds. 📊 The Scale of the Transformation To grasp the magnitude of this transformation, consider the contrast: - The original Intel Pentium chip (released in 1993) had about 3.1 million transistors. - In contrast, one of NVIDIA's latest GPUs, the Blackwell B200, boasts an astonishing 208 billion transistors (67,000x more). With the rise of ultra-high-performance chips and advanced interconnects, this technology is growing exponentially. What was once a single GPU rendering graphics has become part of a vast, layered system: - Multiple GPUs inside a server - Multiple servers connected within a rack - Multiple racks forming clusters (or PODs) - Multiple clusters scaling across entire mega data centers Remember: just one GPU contains 208 billion transistors. Now, imagine the scale of a massive data center built with thousands of them, working together as a single, unified mega computer! 🤯 The result is a MASSIVE AI Factory, capable of training the models we rely on daily and processing the billions of prompts we generate — whether at work or in our personal lives. NVIDIA didn’t just build infrastructure — it redefined the future of large-scale computing and ignited a new era of possibilities for humankind. And we’re only at the beginning of this revolution! 🔗✨ P.S.: We also have the Oracle architecture running in these data centers — but that’s a story for another post. #NVIDIA #Blackwell #B200 #ArtificialIntelligence #DataCenter #DigitalInfrastructure #Supercomputing #AI

AMD has flagged significant risks from Intel and NVIDIA’s planned collaboration to co-develop chips for PCs and data centers, warning in an SEC filing that the partnership could intensify competition and pressure its margins. The alliance aims to combine NVIDIA’s GPU IP with Intel’s advanced packaging and x86 technologies, with first products expected around 2026–2027. Meanwhile, Intel has reportedly lost a key AI executive, Saurabh Kulkarni, to AMD, bolstering AMD’s data center strategy as it targets tens of billions in annual revenue from its Instinct GPU business by 2027. #AMD #Intel #NVIDIA #TechGiants #ChipMaker #Chips #AIChips #Semiconductors #ChipManufacturing #SemiconductorIndustry #DataCenters #TechNews #Innovation #Competition #TechnologyNews https://lnkd.in/dAByZxe2

ASUS Unveils XA NB3I-E12 AI Server Powered by NVIDIA HGX B300 Platform ASUS announced the shipment of the XA NB3I-E12 AI server, built on the NVIDIA HGX B300 platform. Delivering next-generation AI performance and reliability, XA NB3I-E12 gives enterprises and cloud-service providers (CSPs) early access to cutting-edge computing capabilities for the AI era. Accelerated by eight NVIDIA Blackwell Ultra GPUs and dual Intel® Xeon® 6 Scalable processors, ASUS XA NB3I-E12 is engineered for intensive AI workloads. With eight NVIDIA ConnectX-8 InfiniBand SuperNICs , five PCIe® expansion slots, 32 DIMMs, 10 NVMe drives, and dual 10Gb LAN, it transforms data into intelligence for real-world automation. The system is ideal for enterprises and CSPs running large language models (LLMs), research institutions and universities performing scientific computing, and the financial and automotive sectors focused on AI model training and inference. Read More: https://lnkd.in/gfh_rFP8 #ASUS #server #NVIDIA #HGXB300 #CSP #AI #DigitalTerminal ASUS Dinesh Sharma Vaibhav Kulkarni

🎥 Hot Takes | Intel & NVIDIA join forces: A $5B game-changer for AI and data centers 💥 In this exclusive discussion, Yole Group’s John Lorenz and Junko Yoshida break down the strategic $5 billion investment by NVIDIA in Intel Corporation, and what it means for the future of CPUs, GPUs, and AI data centers. Discover how this partnership could reshape the semiconductor landscape, from custom NVLink-enabled CPUs to new opportunities in heterogeneous computing and foundry collaboration. What does this mean for competitors like AMD and hyperscalers designing their own chips? Tune in to find out! 🎙️Featuring: John Lorenz, Senior Technology & Market Analyst, Yole Group Junko Yoshida, Editorial Contributor for Yole Group 👉 Watch the full conversation now on https://lnkd.in/eJEstceS  #Intel #NVIDIA #AI #DataCenter #Semiconductors #YoleGroup #TechAnalysis #marketresearch

Intel Unveils Panther Lake: Ushering in the AI PC Era with Groundbreaking 18A Technology Intel has just revealed its next generation Intel Core Ultra series 3 processor, code-named Panther Lake, built on the most advanced semiconductor process developed and manufactured in the United States—the Intel 18A node. This breakthrough ushers in a new era of AI-enabled computing platforms with unmatched power efficiency and performance. Panther Lake features a scalable multi-chiplet architecture offering enormous flexibility across consumer and commercial AI PCs, gaming machines, and edge devices like robotics. Key highlights include up to 16 performance and efficiency cores delivering over 50% faster CPU performance, a new Intel Arc GPU boosting graphics by 50%, and up to 180 TOPS of AI acceleration. Manufactured at Intel’s state-of-the-art Fab 52 in Chandler, Arizona, Panther Lake will begin high-volume production this year, with wide availability targeted for January 2026. Alongside Panther Lake, Intel’s next-gen server processor, Xeon 6+ Clearwater Forest—also built on 18A—is set to launch in early 2026, delivering substantial gains for data centers with up to 288 efficient cores and a 17% IPC uplift. Intel’s cutting-edge 18A node introduces RibbonFET transistor technology and PowerVia backside power delivery, providing significant improvements in power, performance, and chip density. This generation strengthens Intel’s leadership in U.S. semiconductor manufacturing and innovation as it powers the AI and computing needs of tomorrow. #Intel #PantherLake #AI #Semiconductors #18ANanometer #Intel18A #ChipTechnology #Fab52 #USManufacturing #AIComputing #EdgeComputing #DataCenter #PCInnovation #Robotics #TechLeadership #AIPlatform