Chinese semiconductor thread II

jx191

Junior Member
Registered Member
this is the official statement from the company as they are preparing for the largest ever IPO in coming month. most likely over a Trillion RMB. they are expecting further growth in second quarter and first half revenue will exceed 16-17 Billion dollars (110+ Billion RMB).

CXMT become the first Chinese semiconductor company to enter in 100 Billion RMB revenue club and will also become the first company to open 200 Billion RMB club.
The money is impressive but I think the most important part is the capability and output.

Domestic HBM3 for Huawei's supernodes this year will be huge.

CXMT will play a big role in the domestication of China's AI.

That being said, is this level of revenue and growth sustainable long term?
 

jx191

Junior Member
Registered Member
Did anyone see that report last month about CXMT facing setbacks on HBM3? Taking it with a grain of salt, but they were saying how production of HBM3 at scale could be delayed all the way to 2027. Any thoughts?
 

tokenanalyst

Lieutenant General
Registered Member

BGI NineSky is deploying quantum computing to empower cutting-edge chip design.​


Huada Jiutian has emerged as a key player in China's quantum computing and EDA sectors following its strategic partnership announcement on May 8th with Bose Quantum and other companies. The collaboration aims to expand quantum computing applications across EDA, new energy, and intelligent computing fields, driving increased market interest and stock price growth for the company.

The company has consistently emphasized the critical role of EDA tools in quantum chip design, recognizing that traditional manual and trial-and-error methods are inadequate as qubit counts increase. Quantum EDA (Q-EDA) has become essential as a core software framework supporting quantum hardware modeling, simulation, optimization, and verification, significantly enhancing the design efficiency and accuracy of quantum chips including superconducting and spin qubits.

Beyond quantum applications, quantum computing offers powerful solutions for traditional EDA challenges. Classical computing often reaches bottlenecks when addressing complex NP-hard optimization problems in chip design, such as placement, routing, and new material exploration. Quantum computing's parallel processing capabilities can accelerate these optimization tasks, improving both design efficiency and performance of conventional chips.

China's late start in traditional integrated circuits has resulted in dependence on overseas high-end EDA tools and vulnerability to technological blockades. Industry analysts stress that developing domestic quantum EDA is crucial for achieving self-sufficiency in quantum chip design and manufacturing, while positioning China competitively in the emerging quantum technology era. This proactive approach aims to prevent repeating the dependency issues experienced in traditional semiconductor fields.

Several Chinese quantum technology companies have already prioritized Q-EDA development. Origin Quantum launched China's first industrial quantum chip design software "Origin Kunyuan" in 2022, while QuantumCTek is developing its own EDA software as part of its full-stack quantum computing strategy. These initiatives underscore the strategic importance of domestically produced EDA in China's quantum computing industry chain.

Looking ahead, Huada Jiutian will showcase its quantum computing advancements at the upcoming JIMicro Conference in Shanghai (May 27-29). Senior Director Yu Han is scheduled to present "Beyond Moore's Law - Domestic Q-EDA Full-Process System Empowering Next-Generation Quantum Computing" at the EDA IP Industrial Software forum on May 29th, sharing the company's technological roadmap and strategic vision for quantum computing.​

Please, Log in or Register to view URLs content!
 

tokenanalyst

Lieutenant General
Registered Member

Huawei's Ascend 384 supernode computing cluster has been deployed at Wuxi Hongxin Electronics.​


On May 17, Hongxin Electronics partnered with Wuxi High-tech Zone to establish Jiangsu Province's first Huawei Ascend 384 supernode computing cluster, marking a significant milestone for China's domestic computing infrastructure. Using this cluster as foundational infrastructure, Hongxin Electronics will build a large-scale "Token Factory" in Wuxi—a new model for high-performance computing based on "national chips, national models." The initial deployment includes four Ascend 384 supernode servers, interconnecting 1,536 GPUs into a unified super cluster that breaks away from traditional decentralized computing approaches and delivers ultra-high parallel computing capabilities for large-scale AI model training and inference.

This deployment fully adopts Huawei's end-to-end Ascend technology roadmap, spanning chips, servers, and cluster networking, serving as a practical example of China's push for self-reliant AI infrastructure. The "Token Factory" concept reimagines computing power as an on-demand utility, using tokens—the basic processing unit for large language models—as a standardized metric to deliver cost-effective, scalable computing services. This model emphasizes economies of scale, high resource utilization, and low entry barriers, enabling enterprises, research institutions, and developers to access robust domestic computing power for large-model training, AI inference, and vertical industry applications.

Strategically, this collaboration extends beyond infrastructure construction to explore a viable path for the engineered, commercialized development of domestically produced computing power. As AI development shifts from parameter-centric to application-driven models, demand for stable, efficient, and controllable domestic computing resources is accelerating. Industry analysts view the Hongxin-Wuxi project as a replicable and scalable blueprint for the nation—a "domestic computing cluster + computing service factory" model that can help China's AI industry reduce reliance on overseas technologies and build an independent, controllable computing ecosystem. With Wuxi's strong foundation in integrated circuits and IoT, this initiative positions the region at the forefront of China's high-end domestic computing power development.​

Please, Log in or Register to view URLs content!
 

iewgnem

Captain
Registered Member
Yeah I won’t be surprise if they achieve that. Afterall, the AI boom has been a boon for many semi conductor companies . TSMC has record double digit increase in sales and profits(doubling ) , ASML is making hundreds of billions, japans Kioxia who was facing decline has suddenly gained a new life and just made $8.2billion in profit just this first quarter (yes you read that right ) with even larger sales and expecting a profit of $25billion this year, Samsung market capitalisation has topped $1trillion recently and SK Hynix is about to also reach that figure . I’m actually surprised no Chinese company has reached such a valuation yet. So it should be expected that CXMT should at least reach a hundred billion valuation this coming year. That should be the minimum .
The real implication here isnt the market cap value, its that CXMT is now fully capitalizing on AI memory demand. If you then take a look at every other industry in which Chinese companies gained a foothood, you can get a pretty good idea how everyone else will fair in the coming years
 

sndef888

Captain
Registered Member
this is the official statement from the company as they are preparing for the largest ever IPO in coming month. most likely over a Trillion RMB. they are expecting further growth in second quarter and first half revenue will exceed 16-17 Billion dollars (110+ Billion RMB).

CXMT become the first Chinese semiconductor company to enter in 100 Billion RMB revenue club and will also become the first company to open 200 Billion RMB club.
Anybody with experience in chinese stocks can tell us if it's worth it to try and get into this ipo? Or are the regulations too convoluted / difficult
 

latenlazy

Brigadier
I'm afraid you use "tabloids" level garbage free chatbox, try a paid one.
I’m afraid I don’t use AI to think for me because I can actually think and research for myself. I’m also afraid that no matter how much money you pay for a chatbot LLMs by their design don’t have self vetting functions. They are prone to hallucinations and confirmation bias even if you spend a lot of money on them. Please learn how a tool works before spamming it like a simpleton. The quality of information isn’t strongly correlated with the expense of the toy. Try again.
 
Top