Alibaba has unveiled a new AI chip, Zhenwu M890, which delivers three times the performance of its predecessor. The chip features 144 GB of GPU memory and interchip bandwidth of 800 GB per second. Alibaba aims to reduce its reliance on foreign semiconductor technology, particularly from Nvidia. The company has already supplied 560,000 Zhenwu units to over 400 customers across 20 industries.
Alibaba also announced its next-generation AI model, Qwen3.7-Max. The company plans to upgrade its Zhenwu chip every year, with a focus on AI computing power. This move is part of Alibaba's strategy to invest in domestic AI chip technology and increase its semiconductor capabilities.
Meanwhile, Stability AI has released a new family of audio models, Stability Audio 3.0, which can generate professional-grade music of more than six minutes long. The top model can create full compositions that maintain musical structure and melodic tone.
Anthropic is bringing frontier AI into vulnerability discovery at scale with Project Glasswing, an industry-wide initiative. This move highlights the expanding role of AI in cybersecurity. Google's AI comeback is driven by Gemini, custom TPUs, and strategic cloud investments. The company has introduced Gemini, its most capable AI model, and has integrated it across its product suite.
Nvidia will launch a new research center in Singapore, focusing on embodied AI and increasing the efficiency of AI infrastructure. The city-state is pitching itself as a regional AI hub, ideal for real-world development, testing, and deployment of AI solutions.
Key Takeaways
- Alibaba unveils new AI chip, Zhenwu M890, with 144 GB GPU memory and 800 GB/s interchip bandwidth.
- Alibaba aims to reduce reliance on foreign semiconductor technology, particularly from Nvidia.
- Alibaba has supplied 560,000 Zhenwu units to over 400 customers across 20 industries.
- Alibaba announces next-generation AI model, Qwen3.7-Max.
- Stability AI releases audio model for music generation, Stability Audio 3.0.
- Anthropic launches Project Glasswing for vulnerability discovery at scale.
- Google drives AI comeback with Gemini, custom TPUs, and cloud investments.
- Nvidia to launch research center in Singapore, focusing on embodied AI.
- Singapore to launch testbed for AI robotic technologies.
- Cybersecurity must prepare for AI-driven hardware exploitation.
Alibaba Unveils Powerful New AI Chip
Alibaba has launched a new AI chip called Zhenwu M890, which is three times more powerful than its predecessor. The chip features 144 GB GPU memory and interchip bandwidth of 800 GB per second. Alibaba aims to reduce its reliance on foreign semiconductor technology, particularly from Nvidia. The company has already supplied 560,000 Zhenwu units to over 400 customers across 20 industries. Alibaba also announced its next-generation AI model, Qwen3.7-Max.
Alibaba Unveils New AI Chip for Domestic Use
Alibaba has unveiled a new AI chip, Zhenwu M890, which delivers three times the performance of its predecessor. The chip is designed for AI 'agents' and has 144 GB of GPU memory. Alibaba plans to upgrade its Zhenwu chip every year. The company has shipped over 560,000 units to more than 400 customers across 20 industries. Alibaba also announced a new server system and its latest large language model, Qwen 3.7-Max.
Alibaba's New AI Chip Boosts Performance
Alibaba has unveiled a new AI chip, Zhenwu M890, which is three times more powerful than its predecessor. The chip features 144 GB of GPU memory and inter-chip transfer speeds of up to 800 GB per second. Alibaba has already delivered 560,000 Zhenwu chips to over 400 customers across 20 industries. The company plans to release its next-generation large language model, Qwen3.7-Max, soon.
Alibaba Invests in Domestic AI Chip Technology
Alibaba has unveiled a new AI chip, Zhenwu M890, which delivers three times the performance of its predecessor. The chip is designed to handle heavy memory and communication demands of agent workloads. Alibaba plans to upgrade its Zhenwu chip every year, with a new chip, V900, expected in 2027. The company aims to reduce its reliance on foreign technology and increase its domestic semiconductor capabilities.
Alibaba Accelerates AI Chip Development
Alibaba has unveiled a new AI chip, Zhenwu M890, which delivers three times the performance of its predecessor. The chip features 144 GB of GPU memory and interchip bandwidth of 800 GB per second. Alibaba plans to upgrade its Zhenwu chip every year, with a focus on AI computing power. The company has already delivered 560,000 Zhenwu units to over 400 customers.
Alibaba Unveils AI Chip for Training and Inferencing
Alibaba has unveiled a new AI chip, Zhenwu M890, which is equipped with 144GB of GPU memory. The chip will handle both training and inferencing jobs and is suitable for agentic tasks. Alibaba plans to upgrade its Zhenwu chip every year, accelerating the rate to roughly the same pace as industry leader Nvidia.
New Career Paths Emerge in AI Economy
The AI economy is creating new career paths, emphasizing skilled trades like fiber optic technicians. As AI technologies become more integrated into various industries, the need for individuals who can build, maintain, and operate the underlying infrastructure is growing. This shift is leading to a reevaluation of traditional educational pathways, with a greater emphasis on hands-on skills and vocational training.
Trust3 AI Introduces MCP Security Layer
Trust3 AI has launched Model Context Protocol (MCP) Security, a new standard for safeguarding enterprise agentic AI workloads. The solution forms a key capability within Trust3 AI's enterprise agent control plane, empowering security and governance teams to seamlessly and safely connect AI agents with vital business data, applications, and systems.
AI Transforms Cybersecurity Landscape
Cybersecurity is among the clearest early use cases for applied AI. Anthropic's Project Glasswing, an industry-wide initiative, is bringing frontier AI into vulnerability discovery at scale. The security perimeter is expanding beyond endpoints and networks into code, AI agents, agent identities, private connectivity, and the software supply chain.
Reliability of AI Models Under Scrutiny
The challenge in building reliable AI models is understanding how likely an outcome is and whether we can depend on it repeatedly. Dimensionality and the space of possibilities make it difficult to comprehend the sample space of AI models. The real challenge is not determining whether an outcome can happen; it is understanding how likely that outcome is.
Stability AI Releases Audio Model for Music Generation
Stability AI has released a new family of audio models, Stability Audio 3.0, which can generate professional-grade music of more than six minutes long. The top model can create full compositions that maintain musical structure and melodic tone. The company is making small models available with open weights for anyone to use and modify.
Deferred Context Engine Optimizes AI Performance
Droid's Deferred Context Engine reduces input context by 5.6% on average and reaches a 29.3% p90 reduction. The engine loads internal tool schemas, MCP tool schemas, skills, and plugin instructions only when a task needs them. This approach separates discovery from execution, making AI systems more efficient.
Cybersecurity Must Prepare for AI-Driven Hardware Exploitation
New AI models like Anthropic's Mythos will greatly expand the capabilities of vulnerability researchers to find weaknesses in technology throughout the stack. Cybersecurity must approach this new challenge with a wide view and understand that the entire digital ecosystem will face greatly elevated risks.
Google's AI Comeback: From Lagging to Leading
Google's AI comeback is driven by Gemini, custom TPUs, and strategic cloud investments. The company has introduced Gemini, its most capable AI model, and has integrated it across its product suite. Google's AI strategy involves expanding its AI capabilities and investing in custom hardware.
Nvidia to Launch Singapore Research Hub
Nvidia will launch a new research center in Singapore, focusing on embodied AI and increasing the efficiency of AI infrastructure. The city-state is pitching itself as a regional AI hub, ideal for real-world development, testing, and deployment of AI solutions. Singapore will also launch a testbed for AI robotic technologies later this year.
Sources
- Alibaba Unveils Powerful New AI Chip As China Pushes To Reduce Dependence On Nvidia
- Alibaba unveils new AI chip in push for domestic alternatives
- Alibaba unveils AI chip three times more powerful than previous model
- Alibaba unveils new AI chip in push for domestic alternatives
- Alibaba reveals more powerful Zhenwu AI chip, new LLM
- Alibaba Unveils New AI Chip for Training and Inferencing
- AI Economy's New Career Ladder: The Rise of Skilled Trades
- Trust3 AI focuses on AI agent risks with MCP Security layer
- AI Is Rewriting the Cybersecurity Stack
- From Possible to Probable AI Models
- Stability AI releases a new audio model that can create six-minute songs
- Deferred Context Engine
- Cybersecurity Must Prepare for AI Driven Hardware Exploitation
- Google's AI Comeback: From Lagging to Leading
- Nvidia to launch Singapore research hub as city-state boosts AI plans
Comments
Please log in to post a comment.