pow

Elon Musk stated that the bottleneck in the development of artificial intelligence lies more in power, cooling, and infrastructure, rather than in chips.

Elon Musk stated that SpaceX is "quite confident" about achieving 10 gigawatts of computing power online by 2027; the possibility of exceeding 20 gigawatts of computing power online by 2027 is not zero.

Anthropic's IPO draft shows that the company will invest over $600 billion in computing power.

According to Catcher Predict monitoring, in the Polymarket prediction market for the event "Dota 2: Lynx vs Power Rangers," the sub-market "Day Ends" option's win rate has experienced significant fluctuations, soaring from 50% an hour ago to the current 91% (a fluctuation of 41%). Please note the impact of relevant breaking news.

Bitcoin analyst Sam Callahan stated in a video interview published by Bitcoin Magazine that Bitcoin does not need the Federal Reserve to lower interest rates to succeed; what truly determines its long-term value is fiscal dominance and ongoing currency depreciation, rather than the level of interest rates.Callahan pointed out that, according to the inflation metrics preferred by the Federal Reserve, the dollar has lost at least 20% of its purchasing power over the past five years. He believes that financial repression is already occurring—core PCE averaging 3.7%, and the U.S. bond market is experiencing the steepest pullback in history. He bluntly stated that this policy "works well, but is detrimental to the public."In the interview, he also discussed why the global debt binge makes currency depreciation the easiest way out, the role of the Bank for International Settlements, gold's "monetary protest" and central bank credibility issues, and when central banks will adopt Bitcoin. He listed gold, Bitcoin, and the dollar as the three pillars of reserve assets, believing that all roads ultimately lead to currency depreciation.

Cryptocurrency researcher Kerman Kohli published "Compute Financing" on Substack on October 19, 2026, discussing financing for computing power construction. He believes one of the main bottlenecks is capital, especially credit: early expenditures mainly come from balance sheet cash, followed by a shift to equity financing. The article cites Alphabet's $85 billion financing and Meta's fundraising through shareholder dilution, with the next step being the credit market.Kohli summarizes the current beliefs of creditors into three points, which he believes are incompatible: AI may be a bubble; the biggest risk is counterparty risk in purchase contracts; the underlying GPUs are not valuable and depreciate like ordinary hardware. He argues that demand is closer to infinite, stating that the demand for agentic AI is growing by about 2.5 to 5 times each year, with supply potentially doubling by the end of 2030. Memory is expected to sell out by 2027, with computing power commitments reaching trillions of dollars, and he mentions Jane Street's multi-billion dollar commitments, as well as the needs of pharmaceutical companies, research, military, and cybersecurity.Regarding efficiency, he states that when the cost of intelligence drops by 90%, the increase in usage is even greater, with a baseline exceeding 10 times, and he cites Jevons' Paradox.On the topic of purchasing, Kohli believes that computing power is more substitutable than real estate; when demand is strong, if someone withdraws their commitment, others will take over. He mentions that many creditors require 2 to 3 year contracts and down payments of 20% to 30%, sometimes as high as 50%, and locking in long-term discount contracts effectively shorts computing power for lenders. The article also states that computing power rental prices have risen by about 50% in the past 6 months and about 100% in the past year, with B300 supplies being tight. Regarding collateral, he notes that the A100 from 2020 is still operational and rented out until 2029, leading him to believe it has at least about 9 years of use value; the current price of B200 is about 50% higher than when it was released about two years ago, and a 6-year-old A100 is still worth about 25% of its value. He concludes that computing power will surpass real estate to become the largest asset class, and the scale of credit-supported computing power is currently just beginning.

According to The Information, DeepSeek is collaborating closely with Huawei to jointly design the next generation of AI chips, aiming to build models that do not rely on NVIDIA's computing power. Insiders revealed that Huawei engineers have entered DeepSeek's office to understand its model architecture and algorithm requirements, and to test how the chips handle data transmission and compression, with the first compatible chip targeting Ascend 950DT.The report pointed out that this move comes as China pushes to reduce its dependence on American chips. Limited by TSMC's foundry services and U.S. export controls, Huawei's Ascend 950DT is planned for release in 2027, but mass production may be delayed until 2028. To address the chip shortage, DeepSeek is using the TileLang programming language to bypass CUDA restrictions, while still attempting to acquire the banned NVIDIA Blackwell chips through third-party data centers. In addition, DeepSeek is forming a chip team and has begun testing clusters with over 1,000 processors.

According to Bloomberg, citing informed sources, AI cloud computing company Nscale, supported by Nvidia, is in talks with Anthropic to sell computing power, with a deal value potentially around 45 billion dollars.According to the plan, Nscale will build a second facility at its Monarch data center campus located in West Virginia, USA, providing 460 megawatts of computing power, using Nvidia's Vera Rubin chips. Anthropic also holds a priority purchase right for a third facility. Informed sources said the deal could be finalized in the coming days, but details may still change. According to Nscale's IPO application documents, Anthropic had previously signed an infrastructure agreement worth approximately 44.6 billion dollars with them.

According to a report by the Financial Associated Press, Microsoft has withdrawn from a large artificial intelligence data center project developed by the UK data center company Nscale in Essex, with reports indicating that the withdrawal is related to concerns over power supply. Officials from the UK National Grid had informed Nscale that the power supply for the data center is not expected to be in place until the 2030s. Nscale is exploring whether it can generate power on-site and whether it can expedite the grid connection process.The project was previously referred to by the UK government as the largest data center in the UK, originally scheduled to open next year. As of the report, the site still appears to be just a scaffolding yard. Microsoft had originally planned to fund the construction of a supercomputer, but the collaboration has now been terminated. Following the termination of the related computing power transaction, Nscale lost its only publicly announced client for the data center in Essex. Nscale stated that its relationship with Microsoft remains strong and long-lasting, and the two parties will continue to collaborate on other overseas projects, with Microsoft still participating in another project in Norway.The Starmer government highlighted this project when announcing the UK's 2025 artificial intelligence strategy, calling it the largest autonomous artificial intelligence data center in the UK. Nscale has received support from several tech companies, including Nvidia, and plans to go public in the US this year, with a valuation potentially reaching $35 billion.

Liu Xing, the head of public relations at ByteDance, publicly questioned the claim that "over half of China's AI computing power is used for creating light-hearted dramas," stating that the related data is "somewhat unreliable." He revealed that only about 10% of ByteDance's total computing power is used for training and inference of video generation models, and this percentage also includes model development, not all of it is used for producing short dramas.The controversy originated from an investor communication record disclosed by Chinese Online in June this year. The company stated that 55% of domestic AI Token consumption comes from short dramas and video generation, 24% is used for e-commerce marketing, and software development accounts for only 15%. After this set of data was circulated, it evolved into the claim that "half of China's computing power is used for creating light-hearted dramas," leading to discussions such as "the U.S. creates the future, while China creates light-hearted dramas."

Ethereum co-founder Vitalik Buterin shared that the cost of Bitcoin mining is roughly half derived from capital expenditures on computing power equipment and the other half from electricity consumption; in contrast, the cost of AI inference is almost entirely dominated by capital expenditures on computing power. This difference reflects the distinct cost structures of the two computing businesses. Bitcoin miners need to continuously pay for electricity and determine their profit margins based on mining machine efficiency, coin prices, and network difficulty changes; AI inference, on the other hand, relies more on investments in expensive hardware like GPUs, as well as equipment utilization and computing efficiency.The industrial boundaries of both are also changing. As the demand for AI computing power grows, some Bitcoin mining companies are beginning to shift their existing electricity resources and data center facilities towards AI and high-performance computing businesses, hoping to transition from pure mining to computing infrastructure services. Some mining companies are leveraging previously allocated electricity capacity for mining to enter the AI data center market. However, transformation does not mean that cost pressures disappear: mining operations are more directly exposed to fluctuations in electricity prices and Bitcoin prices, while AI computing businesses must bear risks related to GPU procurement, depreciation, and underutilization. Both types of businesses compete for electricity and computing resources, but ultimately, profitability depends on how these resources are converted into revenue.

According to Bitcoin.com, Bitcoin mining service provider Sazmining has announced the launch of a personal purchase plan, allowing customers to directly own immersion hash boards developed by Block's Bitcoin mining division Proto, which is owned by Jack Dorsey.Sazmining is responsible for the installation, operation, and maintenance of the mining facilities in the United States, and the mined Bitcoin will be sent directly to the customers' personal wallets. Each hash board is priced at $4,187, with a down payment of $1,397, and the remaining amount can be paid in 18 months with zero interest.The equipment is expected to start operating between November and December 2026, with a single hash power of about 200 TH/s and an energy efficiency of about 15 J/TH. Sazmining will charge a uniform rate for electricity and maintenance fees and provide hash board operation services to customers. The company was founded in 2018 and operates mining facilities in the United States, Paraguay, and Norway, serving both individual and institutional clients.

According to Xinhua News Agency, General Secretary of the Communist Party of China and President Xi Jinping emphasized during his inspection in Ulanqab, Inner Mongolia on the 16th that it is necessary to coordinate high-quality development and high-level protection, and to unswervingly follow the path of ecological priority and green development. Xi Jinping learned about the "East Data West Calculation" hub node and the green computing power industry in the Chahar High-tech Zone, and required that completed projects operate efficiently and subsequent construction be completed on schedule.He stated that the computing power network is one of the "six networks" planned and constructed by the country, and the computing power industry is in a stage of accelerated development, emphasizing the importance of "computing power and electricity coordination" and "computing power and water coordination." Xi Jinping also inspected the Huangqi Sea Nature Reserve and required the comprehensive implementation of the Wetland Protection Law.

Analyst Jukan published a discussion on the storage industry's cycle, believing that computing power and storage will tighten or loosen in sync, and that there will not be a situation where computing power is continuously in short supply while storage alone turns to surplus. He also stated that the logic of shorting storage while going long on hard drives does not hold, as hard drives are tied to the storage cycle; if this cycle declines, hard drives will struggle to remain secure on their own.Jukan noted that the market lacks a clear judgment on trends after 2027. Historically, prices often peak one to two quarters before supply increases, followed by an expansion in shipments and a contraction in profit pools. He believes that this cycle is different from those of personal computers, smartphones, and general-purpose servers, as the products used are not standardized, and the triggering factors are also different. Some investors who still view storage through the lens of past experiences of profits during upswings and losses during downswings dominate pricing, leading to a low valuation of storage. He stated that a sharp collapse similar to previous ones is unlikely to reoccur, but a gradual decline is possible; even if Samsung and SK Hynix's profit margins are halved from current levels, the price-to-earnings ratio may still be in the teens, about half of the valuation of hard drive companies.He also believes that when storage prices decline, other demands will absorb supply at various price levels, with local AI-driven personal AI demand expected to emerge. If the storage prices pushed up by hyperscale cloud providers cannot be controlled, local AI will struggle to spread. Referring to the transition from mainframes to personal computers and from cloud to smartphones, he anticipates that this AI cycle will shift from cutting-edge laboratories to local open-source AI, with the latter absorbing storage demand. He simultaneously stated that storage prices will not remain high for over ten years; prices will eventually decline, but new consumer demands driven by local AI and open-source initiatives may provide bottom support.

According to Catcher Predict monitoring, in the market prediction on Polymarket for the event "Dota 2: Power Rangers vs Kalmychata - DreamLeague Eastern Europe Closed Qualifiers Playoffs," the "Yes" option in the sub-market "Ends During the Day" has experienced significant fluctuations in winning probability, soaring from 50% an hour ago to the current 91% (a fluctuation of 41%). Please note the impact of related breaking news.

Box CEO Aaron Levie stated that the significant increase in AI tokens and computing power consumption will come from large-scale autonomous agent clusters operating internally within enterprises, rather than personal assistants. He anticipates that the related tokens and computing power consumption could see an increase of about 100 times.Aaron Levie mentioned that the usage of personal agents will generally be capped per person. Knowledge work is currently limited by the ability of humans to generate, transmit, and produce information; at this stage, most agents are still tasked by humans, compressing about 3 to 5 hours of work into approximately 10 minutes, still constrained by the speed at which humans initiate tasks.He further stated that true significant growth will occur when backend agents are deployed in clusters to execute tasks that are completely decoupled from human initiation. The above remarks were quoted by @MTSlive.

On October 13, Ruantong Power was reported to have received 8,000 NVIDIA GPU cards, specifically the NVIDIA 6000D (domestic "special edition" of the Blackwell architecture), which will be used for the production of related servers.

Multiple venture capital firms in Silicon Valley are competing to secure GPU computing power to provide exclusive resources for their AI startups. a16z established a computing power pool of about 20,000 GPUs as early as 2024 and is negotiating expansion with data center companies Switch and Volta. General Catalyst plans to build a data center for its portfolio companies and has brought in the former head of NVIDIA DGX Cloud in May this year to oversee AI infrastructure operations.Even if AI startups complete their financing, they may not be able to rent GPUs in a timely manner. In April this year, major cloud providers like Microsoft were reported to prioritize allocating computing power to internal teams and large clients, leaving some startups facing months of waiting. a16z previously disclosed that during times of computing power shortages, prices that could originally be secured with a six-month contract might require a three-year commitment. Model training requires a large amount of GPU usage in a short period, while inference demand is difficult to predict, making it easy to waste funds by locking in computing power for several years in advance.a16z is concentrating on purchasing and leasing computing power through the Oxygen project, redistributing it to its portfolio companies, allowing them to access prices and contract terms typically reserved for large clients. In August this year, a16z co-led a $300 million financing round for Volta, which is valued at $2.4 billion, with investments covering both AI startups and data center companies providing computing power. The small venture capital firm Crucible Capital has built a cluster with 576 NVIDIA B300 GPUs for its portfolio companies to test, while also selling some computing power to a cutting-edge AI laboratory. This model allows venture capital firms to take on investments that were previously mainly borne by cloud providers and data center operators.

The computing power and inference cost trading platform Itô co-founder Affaan announced that the Itô team has joined the trading company Jump Trading, responsible for its computing power trading business. Affaan and co-founder Alejandro stated that the team will lead the newly established internal computing power trading desk at Jump Trading and has begun trading in this emerging market.According to their introduction, Itô initially hoped to leverage the new contract market to help enterprises address the centralized AI supply chain risks that the existing system struggles to cover. After communicating with over a thousand funds, enterprises, venture capital institutions, and related practitioners, the team focused on the intermediate market for computing power and inference costs. They noted that the common contract duration for capacity is 3 to 5 years, with about 30% requiring prepayment, while the revenue for late seed round to Series C companies is mostly settled monthly and based on usage, and the secondary market prices for popular chips are also quite volatile.Affaan also mentioned that the total contract value for related medium market clusters sometimes ranges from hundreds of millions to several hundred million dollars. Itô previously built a platform as an end buyer, with team members' backgrounds including Goldman Sachs' credit basket business, open-source engineering tools, and supercomputing cluster research. They stated that Jump Trading's conditions in credit, risk control, trading experience, and hardware led them to choose to advance this business through an internal trading desk.