computing

According to Gate market data, the Hang Seng Index opened lower and then rose sharply this morning. After opening more than 150 points down in the early session, it surged back, recovering losses and turning positive. As of the midday close, the Hang Seng Index rose by 0.07%, reporting at 25,457.33 points, while the Tech Index rose by 0.3%, reporting at 4,790.56 points. The total market turnover reached HKD 132.028 billion, with strong performance in the computing hardware sector. In terms of individual stocks, Cambridge Technology rose by 15.37%, and Shengbang Technology rose by 7.69%.

According to Fortune, Bank of America announced a $250 billion infrastructure investment plan on Wednesday, committing to invest in key infrastructure areas in the United States over the next year. The investment scope covers multiple fields including data centers and computing power, renewable energy generation, energy storage, natural gas, transmission networks, and critical minerals and mining, with the goal of supporting energy security, job growth, and economic competitiveness.

According to market news, the CME Group and Silicon Data will launch computing power futures on October 5.

According to informed sources, Anthropic PBC has reached a deal worth $9.1 billion with cloud computing company Riot Platforms. This deal highlights Anthropic's efforts to meet customer demands and ensure access to sufficient computing resources.Recently, Riot disclosed that the company has signed a 20-year agreement with a business to provide 191 megawatts of computing capacity from its facility in Rockdale, Texas.

NVIDIA CEO Jensen Huang stated that artificial intelligence laboratories and AI startups need computing power.

Lambda, an AI cloud computing service provider supported by Nvidia, is financing $917 million through the leveraged loan market to procure AI chips. As the construction of artificial intelligence infrastructure accelerates, chip financing is becoming a new way for capital investment in the AI industry. Lambda belongs to the rapidly developing "new cloud vendor" camp in recent years, with its main business being to provide GPU computing power and AI infrastructure services to enterprises and developers.This financing plan will be completed through a loan based on GPU asset-related rights, aimed at supporting the company's expansion of AI computing resources. Reports indicate that AI infrastructure companies are actively exploring new financing methods to meet the enormous capital investment required for building large-scale computing clusters. Previously, AI cloud service provider CoreWeave completed the first transaction in the institutional leveraged loan market for chip financing, providing a new financing model for the industry. As the demand for generative AI continues to grow, Nvidia's GPU supply has become a core resource for AI companies' expansion. By using GPU assets as the basis for financing, AI cloud service providers can rapidly scale their computing power without fully relying on equity financing, while also allowing the traditional credit market to participate in the wave of AI infrastructure investment.

The number of registered users on the Zhipu MaaS open platform (i.e., API users) has approached 7 million, an increase of about 2 million compared to early July, including 23,000 enterprise customers. The user count for the developer product ZCode, which competes with Codex, surpassed one million within one month of its launch.Since the beginning of this year, Zhipu's ARR has grown 15 times. It is understood that after the release of new models by Kimi and Alibaba, Zhipu's ARR growth rate has not been affected. Some investors have revealed that the current ARR has reached $2 billion, but Zhipu's official stance has denied this, with people close to the company indicating that the actual figure should be higher. On July 21, it was reported that Zhipu has built a domestic AI computing power infrastructure with a scale of 1GW. It is understood that more than 50,000 domestic computing chips have been activated to alleviate the growing inference demand. On July 31, Zhipu's long-term Coding Plan, which had significantly increased in price, was opened for purchase.

The Token expenditure price index has fallen to 1.1635, continuing to decline since peaking at the end of May (the high of 2.0651 on May 28), with a cumulative drop of 43.7% over two months. This price has fallen below 1.2446 at the end of 2025, officially setting a new low for the year, marking the return of AI computing power pricing to the starting point of this expansion cycle.Note: The sale of Tokens is currently an important source of monetization for the AI industry, and its price is one of the signals measuring the current supply and demand situation for computing power. If the Token price remains high, cloud service providers will be motivated to purchase more GPUs, DRAM memory, and expand data centers; conversely, they need to be cautious of AI companies "downgrading consumption," which could become a leading indicator of the end of this hardware frenzy cycle. Token price data is sourced from top global AI models such as OpenAI, Anthropic, and DeepSeek.

SK Hynix stated in a recent article that the competition in AI is shifting from the performance of individual chips to the design and operation of the entire infrastructure architecture. The competitiveness of AI data centers no longer depends on individual components, but on whether the five key elements of computing, memory, storage, networking, and power cooling can be seamlessly integrated.The article points out that the continuous expansion of AI model scales has led to a surge in demand for computing power and data movement. Training requires repeatedly reading massive datasets, while inference relies on quickly retrieving user request information, both of which place higher demands on the system architecture of data centers. At the memory level, HBM, server DRAM, and others have formed a hierarchical system, each undertaking different bandwidth and capacity tasks. At the networking level, as large-scale training and inference rely on multi-server parallel processing, networking has become a key factor determining the scalability of data centers. System design is shifting from single-server to whole rack and cluster-level expansion.According to Omdia's forecast, the AI data center chip market will grow from $123 billion in 2024 to $207 billion in 2025, reaching $286 billion by 2030. SK Hynix also mentioned that Microsoft's Fairwater data center in Wisconsin is about the length of five football fields, indicating that infrastructure is being deployed on a larger scale. SK Hynix emphasizes that memory is becoming a key layer connecting computing and data.

According to Bloomberg, informed sources revealed that the U.S. government department responsible for investigating violations of chip export controls is currently conducting a systematic review of how Chinese AI companies obtain advanced chips from Nvidia through legal means, focusing on their behavior of renting computing power in third countries. This move stems from recent technological breakthroughs achieved by Chinese AI companies, indicating that they still have the capability to acquire and use top-tier hardware despite U.S. export restrictions on chips to China. Analysts believe that this review may prompt the U.S. to further tighten regulatory rules on computing power leasing in third countries to close existing loopholes in the export control system.

Elon Musk posted that he roughly estimates that about 25% of TeraFab's AI computing output will be used for Tesla Optimus, and about 75% will be used for AI spacecraft.

According to BeInCrypto, since November 2025, Bitcoin's 30-day average hash rate has decreased by 19%, dropping from 1,108 EH/s to 898 EH/s. Glassnode data shows that this nine-month decline is the longest in the network's history.This decline coincides with the largest capital migration in mining companies' history, with mining firms holding over $70 billion in AI contracts, and some of the power capacity originally used for Bitcoin mining has shifted to AI hosting, which may be difficult to return.

According to TechCrunch, AI research startup Mirendil has reached a multi-year partnership agreement with Google Cloud to gain large-scale computing resources to support its "self-improving AI" development.Mirendil co-founder and CEO Benham Neyshabur revealed that the total value of the agreement exceeds $100 million, which is about half of the $1 billion valuation seed round financing amount completed by the company at the end of June.Under the agreement, Mirendil will receive TPU and NVIDIA GPU computing resources provided by Google Cloud, as well as managed AI training clusters, to develop AI systems capable of continuously optimizing their own capabilities.It is reported that Mirendil is dedicated to developing "recursive self-improving" AI, which means AI systems can enhance their performance through continuous iteration, self-learning, and optimization. This direction is also a research area of interest for some of the top AI laboratories currently.

According to official social media news, Huobi HTX will hold a live debate themed "AI Infrastructure Inflection Point: The Ultimate Divergence of Computing Power Bubble and Storage Supercycle" today at 8 PM. At that time, Cong Ge, Jiezou Xia, Cha Ge, and Zhang Facai will represent the affirmative side, engaging in dialogue with the opposing representatives Xiyu, Chan Ge, Bafang, and Wenle, focusing on the core industry variables under the AI wave, and deeply discussing whether the demand for computing power is overestimated, whether there is a bubble risk in AI infrastructure, and whether the storage industry can usher in a new supercycle.

According to Jinshi reports, Musk stated that SpaceX expects its computing power to exceed 2 GW by the end of the year, and by the end of 2027, the computing power may be closer to 10 GW rather than 5 GW.

According to Jinshi reports, SpaceX is collaborating with NVIDIA to design the Starmind AI1 satellite computing payload.Each Starmind satellite will be equipped with NVIDIA Rubin GPUs and Vera CPUs to provide data center-level computing capabilities in space.

According to Jinshi reports, Anthropic has signed a $10 billion computing services agreement with an artificial intelligence cloud startup.

Former hedge fund manager and CNBC host Jim Cramer stated that due to concerns about quantum computing threatening Bitcoin's security, he plans to sell all of his BTC holdings. His statement stems from an interview with IBM Chairman and CEO Arvind Krishna, who said that investors should be wary of the challenges quantum computing may pose to modern cryptography in the next 3 to 4 years. Cramer believes that quantum computing could threaten the Bitcoin network in a similar timeframe. However, no one has independently confirmed how much BTC he holds or whether he has completed the sale.After his statement, Bitcoin continued to trade normally around $63,764, with some market participants viewing his comments as a "reverse Cramer" signal. Bitcoin uses the ECDSA signature mechanism based on the secp256k1 curve, and theoretically, a sufficiently powerful quantum computer could use Shor's algorithm to derive the private key from the public key. The risk is mainly concentrated on addresses with exposed public keys, including reused addresses, early wallet formats, and the brief time window after a transaction is broadcast but not yet confirmed. Researchers estimate that about 6 to 7 million BTC, accounting for approximately 30% of the supply, may fall into this category. Google Quantum AI estimated in March this year that breaking the relevant cryptographic mechanisms could require fewer than 500,000 physical qubits, reducing the previous estimate by about 20 times. However, current quantum systems typically only have hundreds to thousands of physical qubits, with even fewer logical qubits that have higher reliability. Most researchers expect that truly capable quantum computers for cryptographic breaking may not appear until the 2030s or even 2040s, making Cramer's 3-year prediction significantly earlier than most technological expectations.