In light of US sanctions, China unveils first parallel optical computing chip, ‘Meteor-1'
Chinese researchers have developed the first highly parallel optical computing integrated chip, named 'Meteor-1', setting a milestone for using light to perform an enormous number of operations at the same time, the scientists say.
The advance promises hardware acceleration for AI and data centres struggling with soaring
computational demands
The chip achieves a theoretical peak computing power of 2,560 TOPS (tera-operations per second) at 50GHz optical frequency – performance comparable to
Nvidia's advanced GPUs – according to a report by Chinese publisher DeepTech last week.
02:17
Nvidia CEO Jensen Huang unveils plan to build 'AI supercomputer' in Taiwan
Nvidia CEO Jensen Huang unveils plan to build 'AI supercomputer' in Taiwan
Nvidia's latest GeForce RTX 5090 graphic card, for instance, peaks at 3,352 TOPS while its previous flagship RTX 4090 only reached 1,321 TOPS. In the past, optical chips remained mostly in laboratory settings, and could not come close to commercial flagship GPUs in real-life tasks.
Nvidia's 4090 and 5090 are effectively banned for sale to China because of US export controls on advanced semiconductors and AI chips that could aid Beijing in advancing its military capabilities.
As traditional electronic chips hit fundamental physical limits – from heat build-up, quantum effects and unsustainable power consumption – optical computing emerges as a critical future direction. Its inherent advantages, such as ultra-high speed, broad bandwidth, low power and minimal latency, position it to overcome these barriers.
Progress in optical computing has long focused on two key challenges: scaling up the matrix size and increasing optical frequency. Existing top models – exemplified by prototypes from TSMC and the California Institute of Technology – are pushing against both engineering and physical limits.
Consequently, a third way – expanding computational parallelism, or the ability of chips to multitask – has become the necessary path forward.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
The Standard
2 hours ago
- The Standard
Top mainland billionaires see fortunes surge on strength of AI
Zhang Yiming, seen in Palo Alto, California, in March 2020, saw his wealth balloon to over 481 billion yuan following the massive traction of ByteDance's content platforms and its early bet on AI tech, according to Securities Times. Photo by REUTERS
South China Morning Post
3 hours ago
- South China Morning Post
Alibaba Cloud to bring AI capabilities to global data centres this year, executive says
Alibaba Group Holding 's cloud computing unit is set to extend its artificial intelligence (AI) services, including its Qwen family of models, to its overseas data centres this year, according to an executive, as the company seeks to enhance its infrastructure to meet growing demand from foreign companies and domestic clients looking for global opportunities. The company, owner of the South China Morning Post, planned to offer its flagship Qwen large language models, as well as Model Studio – which also provides access to third-party models – through its major data centres worldwide, said Liu Weiguang, senior vice-president of Alibaba Cloud, at an event in Shenzhen on Tuesday. Hangzhou-based Alibaba currently operates 29 data centres, including two in the US, one in London, one in Dubai and 15 across mainland China and Hong Kong, according to its website. Some of the company's AI products are already available in certain international locations, such as Model Studio, which was introduced in Singapore last July. Liu highlighted its global partners Microsoft and Siemens, as well as a partnership with Nvidia in the field of embodied intelligence, which is closely related to robotics. Alibaba planned to secure more partnerships this year, especially in AI and embodied intelligence, he added. 11:13 How is betting on AI to transform e-commerce How is betting on AI to transform e-commerce The initiative is part of an ambitious 380 billion yuan (US$52.4 billion) investment plan announced by Alibaba in February to develop its cloud computing and AI infrastructure over the next three years.
South China Morning Post
8 hours ago
- South China Morning Post
In light of US sanctions, China unveils first parallel optical computing chip, ‘Meteor-1'
Chinese researchers have developed the first highly parallel optical computing integrated chip, named 'Meteor-1', setting a milestone for using light to perform an enormous number of operations at the same time, the scientists say. The advance promises hardware acceleration for AI and data centres struggling with soaring computational demands The chip achieves a theoretical peak computing power of 2,560 TOPS (tera-operations per second) at 50GHz optical frequency – performance comparable to Nvidia's advanced GPUs – according to a report by Chinese publisher DeepTech last week. 02:17 Nvidia CEO Jensen Huang unveils plan to build 'AI supercomputer' in Taiwan Nvidia CEO Jensen Huang unveils plan to build 'AI supercomputer' in Taiwan Nvidia's latest GeForce RTX 5090 graphic card, for instance, peaks at 3,352 TOPS while its previous flagship RTX 4090 only reached 1,321 TOPS. In the past, optical chips remained mostly in laboratory settings, and could not come close to commercial flagship GPUs in real-life tasks. Nvidia's 4090 and 5090 are effectively banned for sale to China because of US export controls on advanced semiconductors and AI chips that could aid Beijing in advancing its military capabilities. As traditional electronic chips hit fundamental physical limits – from heat build-up, quantum effects and unsustainable power consumption – optical computing emerges as a critical future direction. Its inherent advantages, such as ultra-high speed, broad bandwidth, low power and minimal latency, position it to overcome these barriers. Progress in optical computing has long focused on two key challenges: scaling up the matrix size and increasing optical frequency. Existing top models – exemplified by prototypes from TSMC and the California Institute of Technology – are pushing against both engineering and physical limits. Consequently, a third way – expanding computational parallelism, or the ability of chips to multitask – has become the necessary path forward.
