Samsung Q2 Profit Seen Soaring as AI Memory Demand Keeps Chip Prices Elevated 
Wolfspeed Sues Navitas Over GaN and SiC Patent Infringement 
BHP Workers Approve New Labour Agreement at WA Iron Ore Operations 
Netflix, Disney, YouTube Eye FIFA World Cup TV Rights in Multi-Billion Dollar Battle 
AI Memory Chip Shortage Likely to Persist Despite Korea Investment Boom, Nomura Says 
Anthropic Tightens AI Access Controls After Reports of China-Based Workarounds 
SpaceX Stock Draws Bullish Wall Street Coverage Ahead of Nasdaq-100 Inclusion 
Kuaishou Stock Jumps as Kling AI Secures $2 Billion Funding Round 
Meta CEO Zuckerberg Says AI Agent Development Has Slowed Despite Massive AI Investment 
Bank of America Upgrades T-Mobile to Buy, Says LEO Satellite Fears Are Overdone 
Trump Administration to Launch Voluntary AI Standards for Frontier Models 
Barclays Downgrades Siemens Energy as Valuation Seen Near Peak 
China 618 Smartphone Sales Drop 13% as Higher Prices Hurt Demand, Huawei Gains Market Share 
Fiserv Explores Sale of STAR Payments Network as Major U.S. Banks Show Interest 
Super Micro Employees Detained in Taiwan AI Server Export Investigation 
Meta Cloud Ambitions Could Challenge AWS, Azure, and Google Cloud, Says Morgan Stanley 
Knowledge Atlas Tech Joint Stock (HK:2513), the listed company associated with Chinese AI startup Zhipu AI, surged on Tuesday after a report said the company is exploring the development of a custom application-specific integrated circuit (ASIC) to power its GLM family of large language models.
Shares climbed as much as 9.9% in Hong Kong trading, outperforming the Hang Seng Index, which gained about 2%, as investors reacted positively to the potential move into proprietary AI chip development.
According to The Information, Zhipu AI has begun preliminary discussions with several Chinese chip design companies about creating a specialized AI processor tailored to its large language models. However, the company has not selected a development partner, and the project remains in its early stages. The report added that designing the chip could take more than two years.
The initiative comes as demand for Zhipu AI's recently launched GLM-5.2 model continues to surge. Daily token usage reportedly increased 27-fold during the first week following the model's release, significantly increasing computing requirements. The growing demand also highlights the challenges Chinese AI companies face as U.S. export restrictions continue to limit access to Nvidia's most advanced AI chips.
Unlike general-purpose graphics processing units (GPUs), ASICs are built for dedicated workloads, allowing companies to improve efficiency, lower energy consumption and reduce long-term operating costs for AI inference. Developing an in-house processor could also lessen Zhipu AI's dependence on third-party chip suppliers while strengthening control over its AI infrastructure.
If the project moves forward, Zhipu AI would join a growing group of technology companies investing in custom AI silicon. Industry leaders including Google, OpenAI, Alibaba and ByteDance have already developed proprietary AI chips to optimize performance and reduce reliance on external hardware providers. The trend has become increasingly important for Chinese AI firms seeking domestic alternatives amid tightening U.S. technology export controls.
According to the report, the next key milestone will be the selection of a chip design partner, a step that would indicate the project has progressed beyond the exploratory phase.
