pow

Artificial intelligence startup Anthropic disclosed in its confidential IPO filing that it has signed an agreement with SpaceX to pay up to $84.5 billion for AI computing resources based on NVIDIA chips by 2029.The agreement can be terminated with 90 days' notice.

The decentralized AI infrastructure project 0G's computational digital asset Infinite AI (iAI) will open public minting today at 09:00 UTC / 17:00 Beijing time on iai.finance. This is the first landing product of Compute Finance proposed by 0G: transforming staking from "reissuing a bunch of homogeneous tokens" to "a daily spendable AI computing power quota."The issuer has pre-minted the first batch of 2,000 iAI on September 28 according to the same minting curve and locked 0G as collateral for initial liquidity on comfy.fun. Public minting will not open a separate bypass; all will follow the same collateral model.

Noah Shinn, founder of Personal AI Agent Instinct, stated that the company's user base has been growing by about 10% to 11% daily in recent weeks, leading to a corresponding demand for computing power that is close to doubling each week. He now spends about 40% of his time considering how much computing power to purchase in advance.Shinn indicated that computing power cannot be procured in sync with user growth. Purchasing only double the current computing power would be exhausted in about a week; buying five times would run out in less than three weeks; directly purchasing ten times would mean a significant upfront investment. Many resources need to be booked months in advance, and the price for temporarily supplementing computing power could be three to four times higher. He mentioned that purchasing too little makes it difficult to support growth, while purchasing too much would amplify costs when growth slows down.He estimates that even if the daily growth rate drops from 10% to 5% to 8%, after continuous growth for 3 to 4 months, it may still be necessary to prepare computing power for a scale of 100 million users, and the waiting time for new computing power to come online would be roughly the same. He believes that this is more challenging than scaling consumer internet products like Instagram and Facebook, because each new user of the personal agent continuously consumes a large amount of model inference. He stated that Instinct's current scaling speed is even faster than similar situations encountered by Claude Code and Codex in the past.

On September 24, Zach Pandl, the research director of the cryptocurrency asset management company Grayscale, published an article in the column The Stack stating that in the wave of artificial intelligence, the demand and supply paths for the computing power required to train, run, and operate models are diverging. Grayscale believes that this imbalance favors owners of already powered and operational computing capacity and brings growth-oriented investment opportunities.Zach Pandl stated that digital demand can expand instantly, but physical infrastructure such as electricity, data centers, chips, memory, and cloud services takes years to approve, access, and build. When artificial intelligence agents perform multi-step tasks, they may consume 5 to 50 times more compute tokens than typical chatbot interactions, and increased application layer activity will transmit to the underlying computing infrastructure.The article cites data from the International Energy Agency and Lawrence Berkeley National Laboratory, stating that data centers are expected to account for about half of the growth in electricity demand in the United States by 2030, and new projects may take more than five years to connect to the grid. Even if electricity is secured, permits, specialized labor, electrical equipment, cooling systems, GPUs, high-bandwidth memory, and networks are still needed. Continuous value will flow to power producers, data center operators, and artificial intelligence cloud service providers that can convert electricity into computation.

According to Forbes, Bitcoin approached $90,000 earlier this month. Meanwhile, global debt continues to rise, and the "currency depreciation trade" surrounding the decline in purchasing power of fiat currency has become one of the factors driving up assets like Bitcoin and gold.Data from the Institute of International Finance shows that global debt increased by $10 trillion in the first half of this year, surpassing $365 trillion in total. The U.S. debt has exceeded $40 trillion, with annual interest payments rising to $1.27 trillion, surpassing defense and Medicare spending, and only lower than Social Security spending. The institute warns that as benchmark interest rates rise, interest costs will also increase.Nic Puckrin, founder of Coin Bureau and cross-asset analyst, stated that the current environment is favorable for "currency depreciation assets" like Bitcoin and gold, which is part of the reason for Bitcoin's recent rise. The larger the debt of major economies, the more likely it is to suppress real borrowing costs and allow inflation to erode the real value of debt, thereby enhancing the attractiveness of such trades.Analysts from The Kobeissi Letter pointed out that the purchasing power of the dollar has declined by 23% since 2020; if assets only increased by 30% during the same period, investors have essentially just broken even in terms of real purchasing power. The U.S. inflation rate has been above the Federal Reserve's 2% target for 60 consecutive months.

According to Euronews, Kazakhstan is promoting cryptocurrency mining companies to utilize the remaining associated gas from oil fields to build their own power stations, in order to reduce natural gas flaring, alleviate pressure on the national power grid, and increase oil production.Data shows that the country will flare about 300 to 340 million cubic meters of associated gas in 2024, which, if used for power generation, is expected to produce 1.2 to 1.3 terawatt-hours of electricity. Currently, the Ministry of Energy of Kazakhstan and other departments are formulating the relevant legal framework.

Fidelity Investments Global Macro Director Jurrien Timmer posted on X, discussing Bitcoin prices and the power law model.Jurrien Timmer stated: The power law mathematics of Bitcoin continues to indicate that after holding the $60,000 level, a new cyclical bull market is unfolding, with a target of reaching $300,000 by 2029.

A data center developer supported by an affiliate of Blue Owl Capital announced a bond issuance to raise $1.1 billion, betting on AI computing power, to fund the Digital Drive data center project near Richmond, Virginia. This data center has an IT capacity of 76 megawatts and will be fully leased by CoreWeave under a 15-year, $2.94 billion contract, expected to be operational from 2027 to 2028.

According to Bloomberg, a data center developer supported by an affiliate of Blue Owl Capital announced a bond issuance to raise $1.1 billion to bet on AI computing power, funding the Digital Drive data center project near Richmond, Virginia.The data center has an IT capacity of 76 megawatts and will be fully leased by CoreWeave under a 15-year, $2.94 billion contract, expected to be operational from 2027 to 2028.

Market news indicates that Powerlaw is considering a strategic sale of part of its SpaceX shares to raise funds for dividends, tax distributions, and new private market investments.

During an exclusive interview with China Central Television's finance program, Tesla CEO Elon Musk stated that China's AI large models are overall outstanding. He mentioned that, under limited computing power, the output per unit of computing power of Chinese models is already close to the world's top level.Musk assessed that China will solve the computing power bottleneck faster than most people expect. He roughly estimated that in the next 2 to 3 years, China could bridge the computing power gap relying on lithography technology and chip manufacturing capabilities.

According to Grayscale's official announcement, it has renamed its Grayscale Bitcoin Miners ETF to Grayscale AI Compute ETF, with the New York Stock Exchange trading code changed from MNRS to GCPU, and the tracking index changed from the Indxx Bitcoin Miners Index to the Indxx High-Performance Computing Index.The fund will invest in companies that build and operate AI infrastructure, with about half of the holdings in native AI computing companies, and the remainder in digital infrastructure operators that have publicly committed to transitioning to AI businesses, including companies that have transitioned from Bitcoin mining.

Tesla CEO Elon Musk stated that China's AI large models perform excellently overall, with unit computing power output performance nearly at the top globally. He believes that China will solve the computing power issue faster than expected, anticipating that within two to three years, it will close the computing power gap relying on lithography technology and chip manufacturing.

The launch of NVIDIA GPU rental price futures has encountered obstacles. The Chicago Mercantile Exchange originally planned to introduce computing power futures contracts in early October but is expected to be unable to obtain regulatory approval on time.The U.S. Commodity Futures Trading Commission (CFTC) is extending its review to study how the contracts operate and to solicit industry opinions.

Arthur Hayes published a new long article titled "Safety First," with the core argument that the claims of "safety first" by Anthropic, OpenAI, and SpaceX, which lead to a slowdown in AGI development, are not out of concern for human welfare but rather due to economic realities. The market does not want AI; it wants AI at "Chinese prices," meaning it needs intelligence that is 100 times cheaper than what is currently available. Hayes points out that "safety first" essentially destroys the demand for computing power. If the spending on training new models decreases and laboratories shift towards efficiency optimization, customers will spend less on computing power. The three major AI laboratories do not generate any profits, and their demand for computing power supports over $10 trillion in investment-grade debt and hundreds of billions in low-quality debt, which rely on profitable tech companies like Nvidia, Broadcom, Google, and Microsoft for off-balance-sheet endorsements. The real backstop is the holders of insurance policies in the United States.Hayes cites an analysis by Nick Nameth that reveals a "self-insurance scam": private equity giants (such as Apollo, KKR, Brookfield, etc.) acquire insurance companies, stuffing AI data center debt and SaaS private credit impacted by AI into insurance assets, and then provide false endorsements with minimal capital through affiliated self-insurance reinsurance companies. Nameth estimates that the total amount of these false reinsurance assets reaches $1.54 trillion. Once the AI data center debt is downgraded by rating agencies due to insufficient demand for computing power, insurance companies will be forced to add capital, while the affiliated reinsurance companies will be unable to pay, leading to insolvency for the insurance companies. In most states in the U.S., the insurance protection limit is only $250,000 to $300,000, and existing insurance companies only pay into the protection fund afterward, which encourages all parties involved to maximize risk-taking. When AIG was bailed out in 2008, TARP funds ultimately flowed to Goldman Sachs and led to record bonuses, while the general public only received foreclosure notices; Hayes believes this scenario will repeat itself.For cryptocurrency investors, the conclusion is a win-win situation. If the U.S. government chooses to become the "last buyer of computing power," it will print money in the name of national security to fund unproductive economic goods, driving up financial speculation and Bitcoin prices; if the government chooses to bail out insolvent insurance companies, it will also need to print money to cover bad AI debts, increasing the money supply and pushing up Bitcoin. Hayes specifically points out that the Federal Reserve voted unanimously last week to raise interest rates by 25 basis points, and RMP bond purchases have stopped since August 14, but commercial banks have taken over to create over $100 billion in currency, and the interest rate hike allows banks to earn an additional $7.5 billion in excess reserve interest each year. This money will be used to expand loans and market speculation, and the net effect remains stimulative. The fluctuations in the cryptocurrency market, which saw a slight increase at the end of August, are about to end, the supply of dollars will continue to grow, and Bitcoin and some selected altcoins will rise. Hayes also described this situation as "incredibly wonderful," stating that the government will not allow the free market to stop building AI data centers, there will be an oversupply of spot computing power, the usage of AI agents will increase, and the surge in money printing will drive investors to chase cryptocurrency assets.

The general AI computing power chip unicorn EVAS Intelligence announced the completion of a new round of financing, amounting to nearly 2 billion yuan, with a post-investment valuation estimated to be close to 15 billion yuan.This round of financing brought together over 20 investment institutions, including Huatai Innovation, Zhongding Capital, SMIC Juyuan, Jiuan Medical, Renai Capital, Tongfu Microelectronics, Saiyi Industrial Fund, Heli Capital, Xinxin Leasing, Hengxu Capital, Shangqi Capital, Jiangguang Zhanlu, and Rongyi Capital. EVAS Intelligence mainly provides efficient and scalable AI computing acceleration solutions, promoting the widespread application of AI technology across various industries.

On September 16, the day Circle Arc's mainnet went live, the AI Agent quality assurance product DeepQA released the Arc Campaign, completing public QA for 11 ecological applications on the Arc testnet and publishing all 13 reports on the official website in a collection format. From the initiation of the project to the public release of reports for all 11 applications, it took about 16 hours.Real-time data from the official website shows that this event executed a total of 141 test scenarios, of which 117 passed, confirming 15 issues. The Critic Agent conducted real-time reviews 30 times, retaining 141 scenarios, 436 screenshots, approximately 2,200 model calls, and 16.9 million tokens. The covered applications include the Arc official testnet browser, Tower Exchange, AchSwap, Machina Bridge, OmniHub, ArcFX, SettleOne, Syn DAO, thirdweb Arc Testnet, ChainGPT Pad, and Karwan. All tests were conducted in real browsers on live applications, with the injected test wallets holding only testnet funds and no real funds used.The backend of DeepQA is powered by WORLD3 and Google Cloud Vertex AI. A group of AI Agents completed exploration, planning, testing, and reporting, with each issue accompanied by screenshot evidence, and only released after being audited and reviewed in real-time by the Critic Agent. The team stated that a decentralized QA network is also in progress, which will prioritize support for the Arc ecosystem.