pow

On October 29, the cryptocurrency research institution Delphi Digital released a report titled "The AI Boom Has a Power Problem," authored by Accelxr. The report states that deployable AI computing power depends on chip supply, as well as sites that are powered, cooled, and connected to the grid. Electricity is harder to supplement in the short term than chips, as chip production capacity can expand quarterly, while new power generation and transmission take years.The report cites data indicating that the PJM capacity auction has cleared at the price cap for three consecutive times, with the last two still falling short of reliability targets by 6.6 to 6.8 GW, marking the first time such a gap has occurred in the history of this auction. The scale of large load queues in Texas ERCOT increased from 63 GW at the end of 2024 to 474 GW in June, with about 90% being data centers; the governor requested an audit of the data centers in the queue in August, and ERCOT subsequently suspended power approval for data centers and cryptocurrency miners of 75 MW and above, including 17 projects that had passed stability studies.The report also states that U.S. data centers are expected to account for about 4.7% of the nation's electricity consumption in 2024, with LBNL predicting an increase to 9.5% to 15.3% by 2030, and EPRI estimating a maximum of 17%. In a 1 GW facility, chip depreciation is about $2.8 billion to $3.3 billion per year, equivalent to about $8 million per day. SemiAnalysis predicts that by 2028, the U.S. will have over 40 GW of data center capacity utilizing behind-the-meter power sources; the delivery cycle for large gas turbines is about five years, with GE Vernova, Siemens Energy, and Mitsubishi Heavy Industries forming an oligopoly. Wood Mackenzie expects gas turbine prices to reach $600 per kilowatt by the end of 2027, nearly three times the level of 2019.

HSBC assesses that, combined with the previous financing of 35 billion dollars corresponding to over 1 gigawatt, this transaction roughly supports the deployment of 1.7 to 1.8 gigawatts, which still falls short of the management's target of 5 gigawatts by 2027. Future attention needs to be paid to the continuation of funding.

According to Business Insider, AI computing infrastructure startup Hyperbolic has completed a $30 million A-1 round of financing, led by Walden Catalyst, co-founded by Intel CEO Lip-Bu Tan and former Samsung President Young Sohn, with participation from Dell Technologies Capital, General Catalyst, and others.

The United States provides $5 billion in financing to Constellation to enhance nuclear power generation.

On October 21, the payment network Visa announced three capabilities aimed at the growth of commercial cards, the security of virtual cards, and the automation of commercial payments in San Francisco, serving financial institutions, fintech companies, and enterprises. Darren Parslow, Global Head of Visa Commercial Solutions, stated that companies are accelerating the automation of payment-related tasks from identifying vendor opportunities to approving invoices and executing transactions. These capabilities are designed to provide the necessary data, security, and control.The three capabilities are SESA 2.0, Visa Commercial Fraud Solutions, and a live pilot conducted with the billing platform BILL and the accounts receivable platform Billtrust. SESA 2.0 combines AI vendor scoring, Visa commercial payment data, buyer benchmarks, and business value modeling to help identify vendors more likely to accept commercial cards and estimate growth opportunities. Visa claims that since its launch, SESA has helped financial institutions discover over $600 billion in potential cardable spending.Visa Commercial Fraud Solutions covers vendor verification, authenticating commercial virtual card users through 3-D Secure, and transaction-level controls such as spending, merchant category, location, and time. Payment controls have already been provided to eligible issuing institutions in supported countries and regions, while the other two are currently in pilot and are planned for full rollout in 2027. Visa also stated that the pilot with BILL and Billtrust has gone live, allowing trusted AI agents to initiate and assist in completing approved invoice payments through the vendor portal, using tokenized payment credentials without relying on shared vendor portal login credentials.

After Anthropic publicly submitted its prospectus the previous evening, investment institutions and market participants began to assess its growth quality, cost structure, and potential valuation. Morningstar senior equity analyst Malik Ahmed Khan and PitchBook analyst Rolfes stated that as of September 30, Anthropic had over 11,900 enterprise customers contributing at least $100,000 in revenue over the last 12 months, compared to about 950 in the same period last year. Khan believes this reflects a faster adoption of Claude in the enterprise market and significant revenue growth.Analysts also pointed out that the rapid increase in customer numbers alone is insufficient to indicate long-term revenue stability. The prospectus lacks detailed data on customer retention, and the disclosed net revenue retention rate is also affected by a small base; as the comparison base expands, this metric may decline. Investment in computing power is another focus. Morningstar believes that Anthropic currently demonstrates a certain level of computing efficiency and operational efficiency, but some computing resources and related expenditures have not yet been fully reflected; as more computing power is utilized, costs may rise, and if revenue growth does not keep pace, profit margins will be under pressure.PitchBook analyst Rolfes reminded that the scale of future computing contracts and infrastructure commitments is enormous, and whether revenue growth can be sustained is an important variable in assessing profitability prospects. Gavin Baker, managing partner and chief investment officer of Atreides Management, stated that it is still possible for Anthropic to reach an annualized recurring revenue of $100 billion by the end of the year; this statement reflects his personal judgment and is not the company's published performance guidance. Nigel Green, CEO of independent financial advisory firm deVere Group, questioned the market's rumored potential valuation of $2 trillion, calling this expectation "absurd," and reminded to pay attention to infrastructure investments, slowing growth, and future financing needs. The $2 trillion figure remains a market rumor, and the prospectus has yet to determine the offering price and scale.

Arthur Hayes, co-founder of BitMEX and CEO of Flop Labs, proposed in the article "The Trifecta" a plan to establish a spot and derivatives market for AI computing power based on the Flop Network. He believes that there is currently a lack of a unified price benchmark for computing power globally. The Flop Network can form a computing power reference price (FRCR) by recording the amount of computation, time taken, and FLOP payment fees for AI inference tasks.On this basis, users can pre-order computing power for up to 30 days in advance and trade corresponding computing power contracts (FCU), thereby supporting quarterly futures and perpetual contracts settled in USD. Hayes stated that following the concepts of perpetual contracts and synthetic USD stablecoins, he hopes that AI computing power derivatives will become the third innovation widely adopted by financial markets.

Anthropic stated in the S-1 filing that computing power is a key issue the company faces. If it cannot obtain sufficient computing power or fails to allocate it effectively, the company may not be able to train models that remain competitive with its rivals.Anthropic is paying fees to obtain computing capacity from companies like SpaceX, but it expects that demand will continue to exceed supply in the short term.

a16z recently submitted a public comment letter to the U.S. Commodity Futures Trading Commission (CFTC) in response to the CFTC's request for comments on the listing of computing power derivative contracts, expressing support for establishing a regulated computing power derivatives market in the United States. The related derivatives would help AI startups and cloud service providers hedge against computing power price, utilization, and duration risks, and reduce the financing costs of AI infrastructure.a16z stated that the CFTC does not need new statutory authority, as the Commodity Exchange Act (CEA), Part 38 core principles, and Part 40 self-certification process are sufficient, and no new rules are needed prior to listing. The letter suggested that the CFTC rely on the existing regulatory framework to support the adoption of computing power price benchmarks that comply with the principles of the International Organization of Securities Commissions (IOSCO), and allow exchanges to launch cash-settled computing power futures, including fixed-term futures and perpetual contracts.a16z crypto is an investor in Ornn, and ICE announced in May plans to launch GPU computing power futures based on the Ornn Computing Power Price Index (OCPI).

Elon Musk stated that due to the pressure from the debt market environment, over 200 new computing cloud companies will face industry consolidation, with only a few companies able to survive.

According to Slow Fog monitoring, Yapeswap suffered an attack, resulting in a loss of approximately $21,900. The root of the incident lies in the WorkersUnion governance mechanism, where the quorum requirement is zero, allowing a proposal that received only about 0.000055% voting support to transfer the feeToSetter authority of YapeFactory to the attacker.The attacker subsequently deployed a malicious Registry contract, causing latestVault() to return a forged vault address, while vaults() still listed the real vault. YapePair accounted for the deposits to be transferred as farm assets, but did not receive redeemable shares, ultimately leading to a loss of funds. This incident involves contracts such as YapeFactory, YapePair, and YapeWrapper.

According to Catcher Predict's monitoring of on-chain data, there is a prediction of significant capital movement in the Polymarket market. A large whale invested $64,000 in the event "Power Balance: 2026 Midterm Elections" for the outcome "Republican Victory." Based on the current transaction price, the latest corresponding winning probability for this target is 8%.

According to CoinGlass data, the Coinbase Bitcoin premium index has recorded negative values for 31 consecutive days, with the latest at -0.1353%, indicating that purchasing power in the U.S. market has once again fallen into a slump.Previously, on August 24, the Coinbase Bitcoin premium index reappeared with a positive signal for the first time since May 19, reporting at 0.0052%, momentarily ending the longest historical negative premium period of 97 days. The Coinbase Bitcoin premium index is used to measure the price difference of Bitcoin between Coinbase Pro and Binance. A sustained negative index typically means that Coinbase's pricing is relatively low, which may reflect weak buying interest or significant selling pressure in the U.S. market, but it is not advisable to directly conclude that institutional funds are flowing out based solely on this indicator.

Evan Conrad published an article on October 22 titled "The end of startups," with the subtitle "The coming credit crunch." Evan Conrad stated that within the next six months, most AI startups will be unable to purchase computing power, as global computing power is sold out. To obtain capacity, new clusters need to be contracted, and it is becoming increasingly difficult for non-investment-grade entities to secure online capacity.Evan Conrad mentioned that startups are not investment-grade, thus they are required to pay larger upfront payments. He has seen cases where 100% of the contract amount is required to be prepaid, as lenders and cloud providers are unwilling to take on additional credit risk. He believes that investment-grade tenants are scarce, and due to leveraging and selling to non-investment-grade token buyers, their ratings are downgraded, which he refers to as a credit bottleneck. He anticipates that investment-grade GPU buyers will increasingly demand that token buyers themselves be investment-grade or pay large upfront deposits.In the article, he provides an example where, at the time of writing, B300 could be purchased for $6 per hour and then sold for approximately $30 per hour for inference. He expects GPU prices to rise, and the gross margin for inference layers to decline, making it harder for clients with poor credit to obtain tokens. He proposed that lenders treat GPU clusters as collateral and stated that the physical delivery forward market SF Compute could allow lenders to use GPU hours on a liquidity book as collateral, rather than relying on a single service provider's cluster.

The official account of Alibaba Cloud stated that the large model company Step Star has globally released its flagship open-source model Step 5 Preview, positioned as a primary working model for real-world intelligent agent tasks, excelling in AI programming, software engineering, professional knowledge work, and financial scenarios. The model has a total of 600 billion parameters and 27 billion active parameters, utilizing sparse MoE, with a maximum context window of 1 million, capable of processing text, images, and videos. The article mentions that it recently ranked second globally in both weekly and daily call volume on OpenRouter. On the day of the release, Alibaba Cloud announced the completion of DAY0 adaptation, optimized for joint inference based on chips, super node instances, and models, with inference computing power provided by the Lingjun super node.Step Star has developed its own intelligent agent reinforcement learning training platform, with the sandbox platform managed by Alibaba Cloud. The solution is based on Alibaba Cloud's Agent Sandbox and Kubernetes container service, aimed at elastic computing power in training mode, claiming to create tens of thousands of sandboxes in a single minute, with a maximum concurrency of up to one million, compatible with E2B API, and providing native Kubernetes interfaces. Through multi-level image caching, intelligent pre-scheduling, and control link optimization, it achieves millisecond-level application and minute-level creation of 30,000 sandboxes, with the sandbox lifecycle uniformly managed by Alibaba Cloud's backend.In terms of storage, each Pod starts with 15GB of free temporary storage, automatically increasing by 30GB each time when usage exceeds 70%, up to a maximum of 300GB. For security, MicroVM-level isolation is adopted by default, covering computing, networking, storage, and permissions, with zero-trust access to cloud products and large model interfaces. The results provided in the article include: a 90% reduction in image startup time, model evaluation cycles compressed from weekly to daily, and an average savings of about 40% on EBS storage costs; relying on ACK One Fleet, it can support the stable operation of hundreds of thousands of sandboxes simultaneously.

According to CNBC, OpenAI is negotiating a commercial agreement to obtain more computing power from the United Arab Emirates. Investors from the United Arab Emirates and Coatue are leading the financing for OpenAI.

According to Catcher Predict monitoring on-chain data, it predicts significant capital movements in the Polymarket. A large whale invested $86,680.99 in the event "Power Balance: 2026 Midterm Elections," betting on "Democratic Victory." Based on the current transaction price, the latest corresponding winning probability for this target is 67%.

According to The Block, holders of Starknet (STRK) supported a research proposal with over 93.5% of the voting power, which aims to explore the transformation of Starknet from an Ethereum Layer 2 network to an independent Layer 1 network with quantum resistance.StarkWare plans to release a draft white paper in November to further investigate the relevant technological pathways. It is important to note that this vote only supports the initiation of transformation research and does not mean that the transformation plan has been officially approved. Further governance approval will still be required to advance the related transformation.