Qualcomm announced Meta as the first named customer for its Dragonfly C1000 data centre processor at its investor day in New York. The C1000, a general-purpose server chip, won't ship until 2028, making this a forward-looking commitment. Qualcomm also unveiled the AI300 accelerator chip, with the AI200 on track for customer shipments later this year and the AI250 expected in 2027. Separately, Qualcomm confirmed its $3.9 billion all-stock acquisition of AI startup Modular, maker of the Mojo programming language and MAX inference engine — software that enables AI models to run across chips from multiple vendors, directly challenging Nvidia's CUDA lock-in. The strategic play positions Qualcomm as a multi-vendor alternative to Nvidia in AI inference infrastructure, though the company's data centre track record is thin and no benchmarks against Nvidia hardware have been published.
Nguồn: https://thenextweb.com/news/qualcomm-dragonfly-meta-ai-data-center-chips-modular. 8sync News chỉ tóm tắt và dẫn link; bản quyền nội dung thuộc tác giả và nguồn gốc.
OpenAI và Broadcom hợp tác phát triển chip AI tùy chỉnh Jalapeño nhằm cạnh tranh với Nvidia Blackwell và Google TPU, nhắm vào workloads inference. Chip này đã được thử nghiệm với mô hình GPT-5.3-Codex-Spark và dự kiến triển khai vào cuối năm 2025, trong khi tình trạng thiếu hụt HBM đang ảnh hưởng đến biên lợi nhuận của Broadcom.
Lập trình viên nên đọc bài này để hiểu cách các công ty lớn như OpenAI và Broadcom hợp tác phát triển chip AI chuyên dụng, giúp tối ưu hóa hiệu suất cho các mô hình lớn như GPT-5.3, ảnh hưởng trực tiếp đến hiệu năng và chi phí của các ứng dụng AI trong tương lai.
NVIDIA ra mắt NVIDIA Agent Toolkit, một nền tảng mã nguồn mở và mô-đun giúp doanh nghiệp xây dựng các tác nhân AI chuyên biệt đáng tin cậy. Bộ công cụ tích hợp các mô hình Nemotron (tùy chỉnh lý luận), NemoClaw (đảm bảo hành vi an toàn) và OpenShell (thực thi bảo mật), được triển khai trong các lĩnh vực như y tế, an ninh mạng và thiết kế chip.
Lập trình viên chuyên về AI nên đọc bài này để hiểu cách xây dựng các hệ thống agent chuyên dụng, an toàn và có thể kiểm soát được, giúp họ ứng dụng kiến thức về mô hình open-source, bảo mật và tích hợp vào các dự án doanh nghiệp thực tế.

Linux 7.2 is removing two obsolete drivers from its source tree: the DTLK ISA speech synthesizer driver and the Applicom PROFIBUS driver. The Applicom driver was originally ported from SCO Unix in 1998, has had severe coding style issues, was broken for years in early Git history, and has no known active users. Later Applicom hardware only officially supports Windows and has been discontinued. The removals together eliminate approximately 19,500 lines of code, reducing future maintenance burden.
Micron CEO Sanjay Mehrotra has stated that the AI-driven global RAM and NAND shortage will persist through 2027 and only begin to improve gradually in 2028, with no clear timeline for supply catching up to demand. The shortage stems from major chip makers — Micron, Samsung, and SK Hynix — pivoting to serve hyperscaler data center demand, leaving consumer markets undersupplied. Micron posted record Q3 2026 revenue of $41.46 billion, a 346% year-on-year increase. The supply crunch is driving up prices for DDR5 RAM and SSDs, with downstream effects including higher console prices and the Steam Machine launching at $1,049.
Arm-sponsored content arguing that CPUs play a critical but underappreciated role in agentic AI infrastructure. While accelerators handle model performance, CPUs act as the control plane — managing data movement, workload scheduling, and secure isolation. Arm's Neoverse platform underpins custom silicon from AWS (Graviton), Google (Axion), Microsoft (Azure Cobalt), and NVIDIA (Grace Hopper/Blackwell), all reflecting a shift toward purpose-built Arm-based processors in cloud and AI datacenters. The piece introduces the Arm AGI CPU, built with Meta, targeting rack-level density for agentic AI deployments.
AMD's Ryzen AI Halo Developer Platform is a $3,999 mini PC powered by the Ryzen AI Max+ 395 APU with 128GB of unified memory, targeting local AI professionals who need to run massive LLMs without discrete GPU constraints. It can handle 200B parameter models, outpacing even the RTX 5090 in raw model capacity, while undercutting Nvidia's competing DGX Spark (now $4,699) on price. The machine ships with AMD's Ryzen AI Developer Center pre-configured, reducing the historically painful ROCm setup. However, ROCm still lags behind CUDA in maturity — Ollama can still require manual GPU path configuration, and quantization library support arrives later than on CUDA. AMD's upcoming Gorgon Halo platform promises 192GB of unified memory and 300B parameter model support, but closing the software gap with Nvidia remains the key challenge.

The Realtek RTL8723BS WiFi driver, added to the Linux kernel staging area back in 2017 with Linux 4.12, is still undergoing cleanup work nearly a decade later. For Linux 7.2, the staging pull request is dominated by 154 patches targeting this driver, described by Greg Kroah-Hartman as a 'beast of a driver.' The patches focus on removing function wrappers, cleaning up code, renaming functions, and eliminating unnecessary abstractions to bring the driver in line with upstream Linux WiFi driver standards. The RTL8723BS supports 802.11 b/g/n WiFi and Bluetooth 4.0, and the goal remains graduating it from the staging area into the formal networking subsystem.
Orange Pi 6 is a new compact SBC (90x90mm) powered by the CIX P1 (CD8180) 12-core Arm Cortex-A720/A520 SoC with up to 32GB LPDDR5 RAM. Compared to the larger Orange Pi 6 Plus, it features 2.5GbE instead of 5GbE networking, drops LiPo battery support, and comes in a smaller form factor. Key specs include dual M.2 PCIe Gen4 x4 slots, multiple display outputs, a 28.85 TOPS NPU, and support for Debian, Ubuntu, Android, Windows 11, and OpenHarmony. Pricing starts at $239 for the 8GB model, reflecting the high cost of LPDDR5 RAM, making it significantly pricier than typical Orange Pi boards.