gpu

"ReAct (2022, Yao et al.'s paper) is an architecture that makes the LLM 'think before acting' repeatedly."

"IDC: Chinese GPU and AI chipmakers captured ~41% of China's AI server market in 2025, significantly eroding Nvidia's share, which stood at 55% with ~2.2M cards (Reuters)"

""The results were what we expected, which was [that] you’re able to get — out of the same CapEx and OpEx, the same GPUs and energy cost — 6.5 times more, so 550% more, tokens," Bercovici said."

"According to the US federal prosecutor's indictment, Liaw was charged with participating in a plan to circumvent US export controls."

"Chinese semiconductor firms have taken a big chunk of the domestic market, claiming 41% of the local AI server market and delivering 1.65 million AI GPUs out of a total of 4 million units in 2025."

"That is a gap of only +0.0139 PPL, about 0.19%, on the full wiki.test.raw pass (580 chunks, c=512)."

"金正浩指出，AI计算的主导权正加速从GPU向内存转移。"

"The core is AI impacted the video editing industry, the studio laid off employees, and the equipment flowed into the second-hand market."

"“Whether it’s GPUs or CPUs, the demand for AI infrastructure continues to outstrip supply,”"

"MSI will be exhibiting a variety of products, including the NVIDIA GB10 super chip-equipped 'EdgeXpert,' the space-saving 'MS-C926,' the mobility-focused 'MS-C932,' and the compact 'MS-C910E,' optimized for edge AI and industrial applications."

"Nvidia unveiled two initiatives at its recent GTC 2026 event that disrupt the enterprise storage landscape."

"Nvidia launches DLSS 4.5, including a 6x Multi Frame Generation feature exclusive to RTX 50-series GPUs that utilizes a second-gen transformer AI model"

"The reduction in speed is not due to the processing power of the GPU. It's the memory bandwidth."

"This article is the third in a series."

"The biggest concern in the construction and operation of AI data centers is the lag in the development of server equipment cooling systems compared to the development and evolution of AI servers."

""40 cards is a sweet spot, which we found out through various customer surveys," said Li Liu, vice president of Sugon. "32-40 cards can cover most industry needs, while balancing cost and performance.""

"Mistral AI announced on March 30, it has raised 830 million dollars."

"Starcloud Inc., a startup that hopes to build a 5-gigawatt artificial intelligence data center in orbit, has closed a $170 million funding round."

"Multiple institutions, upstream chip manufacturers, and terminal brands have reached a consensus: AI PC hardware prices will be even more intense in the second quarter, and the terminal prices will rise further."

"The company says its software reduces cloud and AI infrastructure costs by as much as 80%."

"Mistral AI has raised $830 million in debt...to fund the purchase of 13,800 Nvidia chips for a major data centre near Paris"

"KubeCon Europe 2026 made AI inference its central focus with major CNCF donations including llm-d, Nvidia's GPU DRA driver and a growing AI conformance program."

"So I'm building it in Zig."

"Jean-Claude Van Anthropic – the only model that can do splits between two GPU clusters 😎"

"This combination of three elements: you (instructing Claude Code on VSCode), Claude Code (writing code), and Google Colab Kernel (executing code on the GPU), allows for the strange experience of 'developing AI with AI'."

"This article explains how to deploy Ollama to Cloud Run (with GPU) on Google Cloud and build an environment where you can talk to the LLM from your local machine."

"I can still find many things that make the Prestige 14 Flip AI+ compelling."

"Red Hat AI Factory is an integrated solution, a collaborative engineering effort with NVIDIA, announced in February 2026."