World's first! China opens a “super high-speed rail” computing network with zero packet loss over 2000 kilometers, helping the East counts and the West counts.

On January 3, 2025, China officially opened the world's first optoelectronic fusion deterministic new computing network infrastructure, marking a major breakthrough in the “East Counts West Counts” project. The facility is built by Jiangsu Future Network Group in cooperation with Zijinshan Laboratory and other units, and the first phase of the project has covered nine cities, including Beijing and Shanghai. This “super high-speed rail” level of computing network has realized a lossless bearing without power relay over 2000 kilometers, with single-wave single-port transmission rate as high as 400G and 800G, jitter of the whole network lower than 5 microseconds, packet loss rate less than one in 100,000, and transmission efficiency of more than 90%.

Liu Yunjie, academician of the Chinese Academy of Engineering, said that the facility will provide a reliable base for artificial intelligence, large model training, etc., and significantly reduce transmission costs. At present, the arithmetic scale of the country's 10 data center clusters has exceeded 1.46 million standard racks, and the green power usage rate of some advanced data centers has reached about 80%. This breakthrough will not only promote China's leadership in global science and technology, but also provide strong support for high-value applications such as smart manufacturing and smart cities.

Harvard Genius Teenager Joins OpenAI, Shortlisted for “Junior Nobel Prize” in High School, Takes a Break from Undergraduate Study to Engage in AI Research

At the beginning of 2025, Jeffrey Wang, a talented Chinese teenager from Harvard University, officially joined OpenAI as a researcher in the basic team, responsible for model pre-training and reasoning. Born in 1995, this talented young man was selected as one of the top 40 finalists for the Junior Nobel Prize in Regenerative Meta-Science for his development of a method for detecting structural changes in 3D genomes when he was in high school.

While at Harvard, Wang taught computer science and statistics courses to 61 students, while also focusing on machine learning, and co-authored papers at top conferences such as NeurIPS and ICML, exploring bias in language modeling privacy attacks and diffusion models. In addition to his academic accomplishments, Wang is also a little-known author with over 6 million cumulative views of his history, science, and statistics articles on the Quora platform.