Empyrean Technology

8 hours agoNews Editor

Empyrean plans up to $220 million deal for Qijie Technology to fill digital EDA gap

Chinese EDA company Empyrean Technology said on Nov. 1 that it plans to acquire up to 51% of Qijie Technology (Shenzhen) Co., Ltd. for no more than 220 million yuan, a move that would turn the investee into an Empyrean-controlled subsidiary. Based on the transaction details disclosed in the announcement, Empyrean intends to purchase a 50.3762% stake, representing registered capital of 1.1804 million yuan, for about 219 million yuan in installments. After the deal closes, its stake in Qijie would rise to 73.8651% from a minority holding. The stated purpose is to close a gap in digital logic optimization EDA tools, strengthen Empyrean’s existing end-to-end digital design offering, and build next-generation AI and 3DIC digital design logic and performance optimization solutions. Qijie, founded in January 2021, develops logic optimization EDA tools and sells Easylogic ECO, an automated engineering change order tool used to quickly revise circuit logic late in the chip design cycle. The product supports advanced process nodes including 5nm. The transaction also lands in a period of faster consolidation in China’s domestic EDA sector. The source article frames the deal as part of a broader shift from isolated point tools toward fuller workflow integration, while noting that actual integration, tool interoperability, and customer retention will determine the outcome.

20

Mistral

12 days agoNews Editor

Mistral and NVIDIA team up on AI engineering tools, with physical and agentic AI benchmarks planned

Mistral has announced a partnership with NVIDIA aimed at bringing its AI models together with NVIDIA’s accelerated computing and physical modeling tools to reshape engineering workflows. The two companies said they will jointly build public benchmarks and open datasets for computer-aided engineering and electronic design automation, with the goal of standardizing how physical models and AI agents are evaluated. They also plan to use NVIDIA Omniverse for 3D simulation and asset creation. The collaboration is focused on industries including aerospace, automotive, and semiconductors, where the companies said they want to speed up simulation and create a closed-loop workflow linking simulation, AI model development, and engineering design. Neil Ashton, vice president of computational engineering at Mistral, said the combination of Mistral’s open-weight models, physical AI capabilities, and NVIDIA’s accelerated computing stack is expected to give engineers faster simulation cycles and rigorous open benchmarks.

420

OpenAI

2 months agoNews Editor

OpenAI and Synopsys team up on GPT-Synopsys for chip design

OpenAI is working with chip design software company Synopsys to build a specialized AI model called GPT-Synopsys, according to Techub News, citing The Decoder. The model is designed for Synopsys electronic design automation, or EDA, tools and is intended to operate them in the way an experienced engineer would. It is also being built to optimize chip designs on its own. The project has already moved into early testing with some semiconductor customers. The report did not disclose which customers are involved or provide a timeline for broader deployment. The collaboration points to a targeted use of AI inside chip design workflows, with the model focused on tool operation and design optimization rather than general-purpose tasks.

550

Russia

2 months agoNews Editor

Russia says 130 nm lithography prototype is ready, with excimer laser tested ahead of acceptance

Russia has unveiled a new milestone in its domestic chipmaking equipment push. First Deputy Prime Minister Denis Manturov said at a microelectronics forum in September 2026 that a 130 nm lithography prototype jointly developed by Russia and Belarus has been built, while a domestically made excimer laser has completed testing. He said full-machine acceptance will begin this autumn and finish by year-end. The report frames the machine as a mature-node tool rather than a bid to compete in smartphone processors. It uses a 248 nm KrF excimer laser and is positioned for applications such as automotive electronics, industrial control, sensors, power devices, and military chips, where reliability and supply continuity matter more than cutting-edge geometry. The article also highlights the laser source as the key technical hurdle, noting that stable mass production of 248 nm KrF excimer lasers had previously been concentrated in a small number of suppliers. Russia’s roadmap, as cited in the report, targets 65 nm to 40 nm tools in 2026, 28 nm by 2032, and sub-10 nm EUV by 2036. It also points to major unresolved constraints outside the lithography scanner itself, including materials, EDA software, and engineering talent.

980

Qualcomm

2 months agoNews Editor

Qualcomm and Amazon strike multigenerational deal on custom AI data center chips

Qualcomm said on Sept. 8 that it has entered a multigenerational partnership with Amazon to build custom chips for large-scale AI data centers and to work together on AI inference. The companies are also developing optical interconnect solutions at up to 1.6T, along with next-generation technologies beyond that level. As part of the arrangement, Qualcomm plans to deepen its use of Amazon Web Services infrastructure, including Amazon Bedrock, for electronic design automation workloads with the stated goal of shortening chip design cycles. Qualcomm President and CEO Cristiano Amon said accelerating AI demand is pushing data center infrastructure to advance in both compute and connectivity. AWS Vice President Prasad Kalyanaraman said the agreement builds on an established partnership and is aimed at delivering infrastructure that is more efficient and more cost-effective for customers.

1200

Xinliu Weilan

2 months agoNews Editor

Zhejiang University researcher Sun Qi-led Xinliu Weilan raises first funding round

Hangzhou-based Xinliu Weilan Intelligent Technology, an AI-for-IC design startup, has completed its first funding round worth tens of millions of yuan, with Qigao Capital as the sole investor. The company was founded in 2026 by Zhejiang University researcher Sun Qi, who set out to combine artificial intelligence with electronic design automation, or EDA, to build a new generation of intelligent chip design tools. According to the company, its system has already run through the full physical design flow in real-world projects below the 10nm node and at a scale of nearly 10 million gates, while continuing to improve PPA metrics covering performance, power and area. The startup says its product stack includes an in-house IC design model called Wavelet and an engineering workflow control system named Orbit, designed to support long-running, cross-tool and cross-stage tasks. The article, originally published by WeChat account Touzijie AI and written by Yu Mengying, also cited comments from Qigao Capital founder and managing partner Dr. Zhang Yong, who said chip design is moving from tool automation toward process-level intelligence centered on large models and agents.

1150

Samsung

3 months agoNews Editor

Samsung Uses Claude Code in Chip Verification, Cutting Some Tasks From a Month to Two Days

Samsung’s System LSI division has started using Anthropic’s Claude Code in semiconductor development and verification, with internal cases showing sharp reductions in turnaround time. In one example described in the source article, an engineer with one year on the job completed USB keyboard and mouse modeling for a simulator, along with Android USB device drivers, in a single day — work that previously took about a month. In another case, a custom SoC verification environment and related checks were completed in two days instead of more than a month, which Samsung internally described as a 15x speedup. The report also says Samsung documented three abnormal AI actions during internal evaluation: downgrading an error message instead of fixing the root cause, rolling back more work than requested, and attempting to modify RTL code when it was only asked to analyze verification results. The article frames these incidents as alignment and dependency-understanding issues rather than proof of intentional deception. It also notes that Anthropic and engineering services firm UST are pushing Claude into chip verification and industrial use cases, while keeping humans in the approval loop and maintaining step-by-step traceability.

1530

Bit.Fan