AMD announced a $1 billion investment in Meta's equity offering to deepen strategic collaboration across CPUs, GPUs, networking, systems, and software, with AMD Helios rack-scale systems already installed at Meta and plans to expand deployment toward gigawatt-scale AI infrastructure.
On October 22, US AI-related stocks suffered a sharp selloff, with the Philadelphia Semiconductor Index falling over 6% and Nvidia, AMD, and Meta among the major decliners. Worries over whether massive AI capital spending will pay off, combined with Middle East tensions pushing oil prices higher, led investors to cut exposure to high-valuation tech assets.
AMD and Intel announced a partnership to optimize PyTorch on x86 CPUs by consolidating their efforts into oneDNN, offering developers a unified software path across both CPU platforms. AMD will integrate ZenDNN into oneDNN and join the UXL Foundation's steering committee, while the companies advance the ACE instruction set for future PyTorch integration.
AMD announced it is bringing its FSR 4.1 upscaling technology to Xbox Series X/S and Ally handheld consoles, using machine learning to enhance image quality and performance. Xbox CTO Scott Van Vilet said Microsoft worked closely with AMD on optimization. The technology will also support RX 6000-series graphics cards and laptops with Ryzen AI 9 processors.
Leaker @金猪升级包 has revealed specs for AMD's unreleased Ryzen AI 7 HX 455 processor, based on the Gorgon Point 1 chip, retaining a 16-core GPU, cutting the CPU to 8 cores/16 threads, with 24MB L3 cache. Its sibling model is the handheld Ryzen Z2 Extreme.
AMD CEO Lisa Su, in a keynote at the OCP Global Summit, characterized 2026 as the "year of inference" and said the CPU market, driven by agentic AI workloads, is expected to exceed $200 billion by 2030 with a compound annual growth rate of about 50%, after AMD raised its forecast three times in the past nine months.
Samsung showcased next-generation memory and storage technologies at the 2026 OCP Global Summit, and demonstrated a CXL-based memory expansion solution jointly with AMD.
On October 12, NVIDIA announced at the OCP Global Summit that it had released Open Data Center Specification v0.7 with AMD, Google and Meta, unifying power, cooling and physical layout designs for AI factories.
On October 12, 2026, AMD announced Helion, a rack-scale computing platform for agentic AI, in Santa Clara, California. Based on OCP specifications and powered by AMD's own chips, it offers a CPU-only configuration with up to 32,768 sixth-generation EPYC cores per rack and a CPU+GPU configuration using EPYC with Instinct MI350P, includes a token router, supports liquid and air cooling, works with existing IT infrastructure, and is expected to launch in November 2026.
AMD has officially launched the next-generation Ryzen AI Halo developer platform based on the Ryzen AI Max+ PRO 495 processor, featuring improved CPU, GPU, and NPU performance, support for up to 192GB LPDDR5X-8533 memory, and the ability to allocate 160GB as shared VRAM for running local models over 300B parameters. The Linux version is priced at $7,299.99, with a Windows version featuring Microsoft Project Zenith to follow.