March 16, 2026 (Local Time) — NVIDIA made a major announcement at the 2026 GTC Conference, launching its full-stack AI infrastructure. The core includes three flagship hardware products: the Vera Rubin Platform, Feynman Forward-looking Architecture, and Groq 3 LPU Inference Chip. Simultaneously, it unveiled the NemoClaw Agent Platform and Dynamo AI Factory Operating System, officially declaring that AI has entered a new era of agents and physical AI.

On the evening of March 12th, AI Chip company Cambricon (688256.SH) released its 2025 annual report, achieving its first-ever full-year profit since its listing. The report shows that in 2025, the company achieved operating revenue of 6.497 billion yuan, a year-on-year increase of 453.21%; net profit attributable to shareholders was 2.059 billion yuan, successfully turning a loss into a profit, compared to a loss of 452 million yuan in the same period of the previous year.

According to South Korean media citing industry sources, Groq has asked the Samsung Electronics foundry (contract chip manufacturing) division to ramp up production. It is reported that Groq recently decided to raise the output of AI Chips outsourced to Samsung's foundry division to approximately 15,000 wafers, up from around 9,000 wafers last year. The production volume last year was primarily used for sample chips to test their suitability for AI inference. Analysis indicates that the company has now entered the initial phase of commercial mass production starting this year.

SambaNova has unveiled its fifth-generation AI Chip SN50, touting it as the only chip capable of delivering the speed and throughput required for agentic AI. It achieves speed up to five times faster than comparable chips, supports a single model with up to 10 trillion parameters and a context window of 10 million tokens through multi-chip interconnect. Compared to Blackwell B200 GPU, SN50 is five times faster at peak performance across a range of models, including Meta's Llama 3.3 70B, and delivers over three times the throughput for agentic inference. For OpenAI's GPT-OSS-120B, SN50 is up to eight times more energy-efficient than B200. SN50 is scheduled to begin shipping in the second half of 2026. On the same day, SambaNova and Intel announced a multi-year strategic collaboration.

Recently, AI Chip company TuringEvo signed a strategic cooperation agreement with the National Integrated Circuit Innovation Center. The two parties will collaborate across the entire value chain in the fields of AI computing chips and key core chips, focusing on chip design, process adaptation, supply chain collaboration, and industrial implementation.

SHENZHEN, February 12, 2026 — The Shenzhen Municipal Bureau of Industry and Information Technology has issued the “Artificial Intelligence+ Advanced Manufacturing Action Plan (2026–2027)”, marking a strategic shift in the city’s AI and semiconductor policy. For the first time, the plan officially supports emerging chip architectures such as in-memory computing and compute-in-memory (CIM) processors, targeting AI applications in smartphones, intelligent robots, and autonomous vehicles. It emphasizes the development of high-performance SoC control chips and promotes domestic alternatives in core AI components.

According to South Korean media outlet ZDNetKorea, global semiconductor giant Samsung Electronics is considering raising prices for some advanced process technologies in its foundry business, primarily targeting the currently high-demand 4nm and 8nm processes, with an average increase of around 10%.

At a recent "trillion-dollar banquet," NVIDIA CEO Jensen Huang revealed that the company is in full-scale production of its next-generation AI Chips, Grace Blackwell and Vera Rubin, and TSMC's production capacity is expected to more than double over the next ten years.

SK hynix Inc. announced that it will open a customer exhibition booth at Venetian Expo and showcase its next-generation AI memory solutions at CES 2026, in Las Vegas from January 6 to 9 (local time).

Up to $3 Billion Deal! Nvidia Reportedly in Talks to Acquire Israeli AI Startup AI21 Labs

Behind the $20 Billion Non-Exclusive Deal: NVIDIA Acquires Groq's Core Team to Reduce Reliance on HBM

According to Korean media reports, Samsung Electronics has recently completed quality testing of its 12-layer HBM3E products with Broadcom and is now negotiating mass production supply. The current discussions involve an estimated supply volume of approximately 1 billion gigabits (Gb), with mass production expected to begin as early as the second half of this year and extend into next year.

During its first-quarter earnings call with analysts, Kim Jae-joon, executive vice president of Samsung’s memory business, said the company is already collaborating with multiple customers on custom versions based on both HBM4 and the enhanced HBM4E.

Samsung Electronics posted KRW 79.14 trillion in consolidated revenue, an all-time quarterly high, on the back of strong sales of flagship Galaxy S25 smartphones and high-value-added products. Operating profit increased to KRW 6.7 trillion despite headwinds for the DS Division, which experienced a decrease in quarterly revenue.

In addition to advancements in custom chip design, significant changes are also occurring in the memory market.