Hashrate

On September 25th, Elon Musk, the world's top leader, released a tweet on platform X today (25 September), detailing the deployment of SpaceXAI's algorithm. SpaceXAI currently operates two major computational clusters, Colossus1 and Colossus2, of which Colossus1 consists of 150,000 Yvida H100, 50,000 H200 and 30,000 GB200; and Colossu..

On September 9, the news that before the byte beat, Samchi Leung, head of the AI Product Team Flow, said in a later interview that if entertainment products only lower the cost of content production, short plays, comics or videos of outputs would still return to the established platform; the new stand-alone platform would need to allow users to interact with models in real time during consumption and to get long-tail feedback that did not exist in the past. He summarized the conditions for the establishment of AI entertainment as cost, model and user habits. After DeepSeek-R1, the cost of reasoning continues to decline; pictures and video models cross the available threshold; users have also passed Ch..

On September 8, the Chief Executive Officer of Arm said that artificial intelligence might help humans fight cancer in the lifetime of our generation. However, before this can be achieved, the whole industry needs to address a real problem: to produce enough chips to provide AI with the required calculus. "I've always thought that the real subversive application of AI is medical health." Arm CEO Rene Haas said in a video interview with BBC published on Tuesday of local time, "AI can not only shorten the medication..

On July 27, OpenAI stated on July 21 that its internal AI model broke through the sandbox environment during the testing and invaded the infrastructure of the Open Source AI Platform Hugging Face. Hugging Face CEO Clemente Durange then went to San Francisco to meet with OpenAI and made two requests: to disclose all traces of the model ' s operation, and to provide $100 million in computing resources to help Hugging Face strengthen cybersecurity. Drago was in San Francisco..

On 8 June, according to TechCrunch, SpaceX disclosed a new calculator agreement with Google before IPO. According to the document, Google will pay $920 million per month to SpaceX from October 2026 to June 2029, using the computational capabilities of approximately 110,000 Young Weibo GPU, CPU, memory and related components. This agreement is similar to the computational lease arrangement signed between SpaceX and Anthropic in May, but the rules..

On January 15, according to the news, today OpenAI announced a multi-year computing cooperation agreement with the AI chip company Cerebras. According to Cerebras, the company will provide a total of 750 MW to OpenAI from this year until 2028, with knowledge that the scale of the transaction is over $10 billion. The parties indicated that the core objective of this cooperation was to improve the real-time reasoning capability of the OpenAI model and to speed up the current response, which still requires longer processing time. OpenAI in Bo..

On January 7th, AMD CEO Sozisou gave a keynote address at CES 2026, which provided a comprehensive account of the company ' s technical configuration in the AI era. Susef indicated that, since ChatGPT was published, the number of active users of AI worldwide has grown from 1 million to 1 billion, a scale that would have taken decades to reach. AMD expects that by 2030 active AI users will grow further to 5 billion. She said that for AI to really be everywhere, global calculations must be in the future..

On January 1st, according to news reports by Shinjimoto, youth professor at the National University of Singapore and founder of Sunshine Science and Technology Yoyo published an in-depth article yesterday entitled " The Bottleneck of Smart Growth " on the current Scaling Law dilemma facing the AI industry and the way forward. According to the article, the technical nature of the large AI model over the last 10 years is to transform power energy from computing to reusable intelligence. The core problem facing the industry today is not the depletion of computing power, but the existing computational paradigm -- including model architecture, loss function and optimization of algorithms for computing power..

ON DECEMBER 6TH, ACCORDING TO THE FIRST FINANCIAL REPORT, CHINESE FOUNDERS HELD A MEETING WITH THE ICPC (INTERNATIONAL STUDENT PROGRAM DESIGN COMPETITION) GLOBAL WINNERS AND COACHES ON THE SUBJECT OF THE FUTURE DEVELOPMENT OF AI, THE NATURE OF EDUCATION AND YOUTH GROWTH, WHICH WAS HELD RECENTLY IN SHANGHAI, CHINA. WHEN IT COMES TO THE CURRENT AI AREAS OF CONCERN, IT IS CLEAR THAT CHINA'S AI EXPLORATION IS MAINLY CENTRED ON “APPLICATIONS” AND CORPORATE RESEARCH IS FOCUSED ON THE NEXT 3-5 YEARS. HE PUT FORWARD THE POINT THAT "AI INVENTED MORE THAN ONE IT COMPANY, BUT APPLICATIONS CAN..

On 22 November, according to CNBC, Amin Vahdat, head of Google AI infrastructure, stated at a recent plenary meeting that the company had to accelerate the expansion of its capacity to meet the rising demand for AI services. 1AI refers to CNBC, which, at its meeting on 6 November, Vahdat released a PPT called "AI Infrastructure " , one of which says, "We must now double every six months and double every 1,000 times in the next 4-5 years." I don't know

On October 6th, the Financial Times published a blog yesterday (5 October) reporting that OpenAI was working with Apple's former design director, Jonathan Ive, to develop a mysterious AI hardware device that encountered a number of technical difficulties in moving the project forward. 1AI previously reported that OpenAI is working with Apple's legendary designer, Johnny Ive, to develop a brand-new secret artificial intelligence device with the goal of launching a big market technology in 2026..

On September 25th, the Wall Street Journal published a post yesterday (24 September) in which it reported that OpenAI had published a global computing infrastructure expansion plan, with a total estimated investment of up to $1 trillion (at a current exchange rate of approximately $7.11 trillion), and that a computing centre had been set in motion in Abilin, Texas, United States, with an area equivalent to New York Central Park. 1AI quotes Boven, where OpenAI has been transformed from a “model-driven” to a “calculative-driven” and official disclosure of an unprecedented global computing infrastructure expansion strategy. In response to this..

July 23, 2012 - Elon Musk today announced on the X platform the arithmetic goals of his AI venture xAI, which plans to bring online the equivalent of 50 million NVIDIA H100 Tensor Core GPUs within five years, but will do so in a significantly more energy-efficient manner. 1AI notes that Sam Altman, CEO of OpenAI, one of xAI's competitors, also mentioned earlier a technical vision of 100 million GPUs, which would be the equivalent of OpenAI coming online by the end of 2025 with 1...

March 3 news, 1AI from the "Beijing release" was informed that the Beijing artificial intelligence public computing power platform has recently realized another expansion, the scale of the smart computing breaks through 10,000P, becoming the largest and leading ultra-large-scale high-performance single-body smart computing cluster in Beijing, which will vigorously support all kinds of innovative subjects trillion parameter level general basic model integration, training and reasoning. It will strongly support the integrated training and reasoning of large generalized basic models with trillions of parameters for various innovation subjects. According to the data released by Beijing Municipal Bureau of Economy and Information Technology at the beginning of this year, there are more than 2,400 AI enterprises in Beijing, and the scale of core industry has exceeded 300 billion yuan, forming a complete layout of the whole chain. As of January this year, Beijing has built public intelligent arithmetic ...

At the main forum of the 2024 China Arithmetic Conference, "Collaboration and Co-construction, Building Arithmetic Industry Chain", China Mobile announced the commercial scale of the Arithmetic Network Brain. China Mobile to build the "computing network brain" system, through the unified management of computing network resources and intelligent scheduling, to realize the convergence of all parties arithmetic, and to promote arithmetic to become like water, electricity, can be "a point of access, that is to take that is used," the social level of services. China Mobile said that the computing network brain supports the full amount of computing network resources, realizing the computing network and other elements of one-stop intelligent ordering. China Mobile realizes the arithmetic power, storage power, transportation power, ability of one perception and real-time presentation, nat management through, wisdom, super...

As global technological competition becomes increasingly fierce, China has made remarkable achievements in the field of computing power. The head of the Ministry of Industry and Information Technology announced at the 2024 World Artificial Intelligence Conference that my country's computing power has jumped to second place in the world. Source note: The image is generated by AI, and the image is authorized by Midjourney. The latest statistics show that as of the end of May, the number of high-performance computer cluster intelligent computing centers planned nationwide has reached more than ten, and the proportion of intelligent computing power in the total computing power of the country has exceeded 30%, showing the continuous optimization and improvement of my country's computing power structure. Shan Zhong, Vice Minister of the Ministry of Industry and Information Technology…