NAll news

All news

Tuesday184 items

  1. Wallstreetcn
    AI score82

    Bank of America, Blackstone, Citi and Morgan Stanley Jointly Provide $63 Billion in Compute Financing for Anthropic

    Bank of America, Blackstone, Citi and Morgan Stanley are providing a total of $63 billion in financing through a newly established SPV, to be used to purchase Broadcom XPU AI accelerators and networking equipment and lease them to Anthropic, with chip deliveries expected to begin in 2027. Broadcom provides the SPV with residual value support equal to the size of the first-lien financing, effectively a guarantee for senior creditors. The transaction is part of Broadcom's strategic compute agreement with Anthropic; previously, Apollo and Blackstone had reached a financing arrangement for a $35 billion compute expansion.

    #Industry news#English text is being prepared.#English text is being prepared.

  2. Wallstreetcn
    AI score62

    Kimi K3.1 Pro Apologizes 48 Hours After Launch, Compensates Subscribers with 25% Usage

    Moonshot AI has apologized for the rate limiting and abnormal quota consumption encountered by subscribers after the launch of Kimi K3.1 Pro, and will provide subscribers within the validity period with a usage compensation of 25% of their monthly total; accounts with less than 25% quota remaining will be reset directly to 0. Some users on the 199 yuan tier said that merely installing one Skill consumed 70% of the 5-hour rolling limit, and the company simultaneously lowered its API per-token pricing. The full weights of K3.1 Pro will be released before November 4; until then, rate-limiting measures will continue.

    Same story · 7 reports

    #Industry news#Agents#DeepSeek

  3. CNBC Technology
    AI score85

    Anthropic Files for IPO, Potentially the Largest Listing in History

    Anthropic filed for an IPO on Monday, potentially the largest initial public offering in history, with investors already valuing the five-year-old company at trillions of dollars. Its second-quarter revenue rose to $11.56 billion from $787 million a year earlier, with losses exceeding $80 billion in the first half of the year and nearly $132 billion in cash and short-term investments as of the end of June. The article notes that nearly half of Anthropic's revenue comes from selling Claude through the cloud divisions of Microsoft, Amazon and Alphabet; in the IPO it may issue about $77 billion in notes to Amazon, Broadcom and Google, while Microsoft and Nvidia have agreed to invest $15 billion.

    Same story · 128 reports

    #Industry news#OpenAI#Anthropic

  4. Shanghai Observer
    AI score58

    China and the US Hold AI Protein Design Competitions Concurrently; Yangtze River Cup Expands Wet-Lab Scale to 7,000

    After the protein design challenge co-hosted by Anthropic and Adaptyv excluded Chinese residents and institutions in China, Shanghai launched the globally open "Yangtze River Cup" AI Protein Design Challenge within a week, with registered teams surpassing 1,200. Both competitions pose problems around 5 health-related targets, offer at least 5,000 wet-lab opportunities free of charge, and will publicly release all experimental data, including results that fall short of expectations. Thanks to increased support, the "Yangtze River Cup" has expanded its wet-lab validation scale from 5,000 to 7,000 candidate proteins.

    #Industry news#English text is being prepared.#Open-source ecosystem

  5. InfoQ AI
    AI score62

    AGNTCon + MCPCon 2026: Agentic AI Foundation Projects Grow to 10…

    At AGNTCon + MCPCon North America 2026 held in San Jose, the Agentic AI Foundation added four new projects, growing from three founding projects to ten, and the agenda's focus also shifted from new protocol features to security and identity. Alibaba contributed OpenSandbox, OpenAI contributed Open Responses, Anthropic joined with Agent Skills, and Hugging Face joined with OpenEnv. Anthropic's MCP co-creator David Soria Parra announced that Claude Code supports MCP tasks, with tool calls that can last from minutes to days; OpenAI said that about three-quarters of the tool schemas on its published servers are smaller than 1 kB, and that every tool must carry annotations to be accepted.

    #Industry news#Agents#English text is being prepared.

  6. Hacker News Trending
    AI score72

    Cloudflare discloses GLM-5.3 Workers AI cross-request KV cache leak and completes fix

    Cloudflare disclosed that the GLM-5.3 Workers AI model had a cross-request KV cache leak; from August 27 to September 9, a total of 155 responses were affected, all involving long prompts exceeding 681,280 tokens. The cause was that the prefill KV pool held only 681,280 tokens while the context window was 1,048,576 tokens; combined with a flaw in SGLang's handling of overly long requests in disaggregated prefill/decode mode, decode nodes read old cache pages that still contained other users' data. Cloudflare has expanded the prefill pool to about 2.04 million tokens and patched the engine; the relevant fixes entered SGLang v0.5.20 through v0.5.22, and operators of disaggregated deployments are advised to upgrade to v0.5.22 or later.

    #Safety / alignment#English text is being prepared.#Open-source ecosystem

  7. Robotics & Automation News
    AI score60

    Qualcomm Announces Acquisition of PickNik to Expand Open Source Robotics and Physical AI Capabilities

    Qualcomm announced that it has agreed to acquire robotics software company PickNik, which has long maintained the open source robot manipulation framework MoveIt. Qualcomm stated that MoveIt will continue to be open source under its existing license, and it plans to integrate MoveIt and PickNik's commercial platform MoveIt Pro more closely into the Dragonwing robotics platform, while also advancing integration with the Arduino VENTUNO Q development board. The transaction remains subject to customary closing conditions.

    #Industry news#Open-source ecosystem#Embodied AI

  8. SiliconANGLE
    AI score76

    Meta plans to offer up to $95 billion in stock to finance AI data center construction

    Meta announced it will offer up to $95 billion in stock to finance data center projects, with the first round of transactions totaling $35 billion, about two-thirds of which comes from common stock, while NVIDIA, AMD and Broadcom have expressed interest in subscribing to up to $5 billion. The company will also issue $12.5 billion in Series A depositary shares, each representing one-twentieth of a Series A mandatory convertible preferred share, and these securities will convert into Meta common stock on August 15, 2029. Meta subsequently plans to raise up to an additional $60 billion through an at-the-market offering, with issuance beginning early next year. The funds will be used for four AI superclusters, among which the Hyperion campus in Louisiana is expected to provide 5 gigawatts of computing power, and Meta will also build 10 new power plants to meet its energy needs.

    Same story · 17 reports

    #Industry news#English text is being prepared.#OpenAI

  9. Unite.AI
    AI score94

    OpenAI Says Its Model Produced Results in Three Research Areas: Mathematics, Computer Science, and Physics

    On October 26, 2026, OpenAI announced that its internal model had produced three research results in the fields of mathematics, computer science, and physics, including a counterexample to the long-standing Exponential Time Hypothesis (ETH) in computational complexity theory. In mathematics, the model provided a proof of the local smoothing conjecture in arbitrary dimensions, and thereby resolved the Kakeya conjecture and the Stein restriction conjecture; the output was obtained on October 18 and Lean formalization was completed the next day. The computer science result was verified by a group led by Russell Impagliazzo, Ramamohan Paturi, and Ryan Williams, and OpenAI agreed to disclose all raw data and fund related conferences; the physics result provided a rigorous existence proof for a four-dimensional quantum field theory with gauge fields and interacting fermions, and also included a pure Yang–Mills existence proof, with the group led by David Gross deciding to temporarily not disclose the AI-produced materials and to write up the results themselves.

    Same story · 17 reports

    #Papers/Research#Reasoning#OpenAI

  10. CNET
    AI score69

    Microsoft offers voluntary severance to about 60,000 employees, while ramping up AI infrastructure investment

    Microsoft announced that it is offering voluntary severance to about 60,000 employees at level 70 and below in the United States, Canada, India and the United Kingdom, roughly a quarter of its total workforce; the company did not say how many would ultimately accept. Microsoft previously disclosed that it would invest about $116 billion in property and equipment in fiscal 2026, mainly for cloud and AI data centers, about $50 billion more than in fiscal 2025. According to Bloomberg, the program covers teams including Xbox, Windows, Copilot and finance, but excludes some data center, sales and marketing personnel.

    Same story · 15 reports

    #Industry news#Microsoft

  11. BleepingComputer
    AI score74

    Former South Korean Penetration Tester Arrested for Stealing 8.2 Million Records Using ARTEX AI

    Seoul police arrested 7 suspects, among whom the main culprit, who had previously worked as a penetration tester, used the ARTEX AI penetration testing framework to attack more than 90 organizations and steal 8.2 million records. In 38 incidents, the suspect demanded 360 million won (approximately $270,000) in ransom in exchange for not leaking the data. Police said this is South Korea's first case in which a suspect was arrested for using generative AI to carry out large-scale hacking and data theft, and noted that AI has lowered the threshold for cybercrime.

    #Safety / alignment#Agents#Industry news

  12. LMSYS BlogFirst-party
    AI score58

    SGLang discloses cross-request KV cache leak and releases v0.5.21 fix

    SGLang officially disclosed that Cloudflare reported on September 9, 2026 that its GLM-5.3 deployment on Workers AI had a cross-request KV cache leak. The cause was that in the PD-disaggregated deployment, the prefill pool capacity of 681,280 tokens was smaller than the configured 1,048,576-token context length, so rejected requests still entered the handoff process and read the caches of other sessions. Upstream fixes #36988 and #38935 make rejected requests return an error before entering the disaggregated handoff, and introduce a delayed-release mechanism that releases slots only after receiving ABORT_ACK from each prefill rank or after a timeout. v0.5.21 now protects Mooncake and NIXL deployments by default, and MORI support and related metrics will be released with v0.5.22.

    An aborted request's launched writes land in KV cache slots that have already been reassigned to another request.

    #Safety / alignment#English text is being prepared.#Open-source ecosystem

  13. East Money
    AI score70

    Meta plans to raise $95 billion through equity financing to invest in AI infrastructure; Nvidia, AMD, and Broadcom intend to subscribe up to $5 billion

    Meta disclosed in a regulatory filing that it has launched a public equity offering of approximately $35 billion and plans to raise up to an additional $60 billion through an at-the-market offering of non-voting Class C shares, to fund investments in AI infrastructure and personal superintelligence. Nvidia, AMD, and Broadcom have expressed intent to collectively subscribe up to $5 billion of Class A stock in this offering, but the intent is non-binding. Meta expects 2026 capital expenditures to be between $135 billion and $145 billion, and the $60 billion at-the-market offering is expected to launch no earlier than the first quarter of 2027.

    Same story · 17 reports

    #Industry news#English text is being prepared.#OpenAI

  14. East Money
    AI score88

    Anthropic Submits S-1 Filing to SEC, Officially Launches IPO

    On October 26, Anthropic submitted an S-1 registration statement to the U.S. Securities and Exchange Commission, officially initiating the IPO process. It plans to list on the Nasdaq Global Select Market under the ticker symbol ANT; the number of shares to be issued, the price range, and the listing date have not yet been determined. The prospectus shows that the company's revenue in the first half of 2026 was $16.286 billion, with a GAAP net loss of $80.748 billion, of which $72.388 billion came from fair value changes in convertible notes, derivative liabilities, and SAFEs; on a non-GAAP basis, adjusted operating profit for the first half was $1.618 billion. Founder LLC, composed of 7 co-founders, holds Class F shares carrying 50.1% of voting power, while the Class A common stock held by public investors carries only one vote per share.

    Same story · 128 reports

    #Industry news#OpenAI#Anthropic

  15. Wired AI
    AI score69

    Anthropic IPO Prospectus Omits Multiple Key Financial Figures

    Anthropic filed its IPO prospectus S-1 on Monday, disclosing its business, strategy and risks, but omitting revenue figures for the most recent quarter (ended September) and details of lease agreements related to compute investment. A footnote in the prospectus shows that since December 31, 2025, the company's future minimum payment obligations under compute and infrastructure leases it has initiated or amended total approximately $417 billion, and if all agreements are fulfilled, it would bear a minimum of approximately $808 billion. Several scholars interviewed noted that the omitted information may widen the information gap with outside investors and could affect IPO pricing.

    Same story · 128 reports

    #Industry news#OpenAI#Anthropic

  16. Yicai
    AI score55

    AI Spending Surge Drives Record Annual US Investment-Grade Bond Issuance

    Driven by a surge in artificial intelligence spending that has pushed companies to take on more debt, US investment-grade bond issuance this year is expected to set a historical record on Monday, surpassing 2020's $1.75 trillion, with year-to-date issuance reaching $1.76 trillion once Monday's trades are priced. Since 2026, AI-related financing in the global high-grade bond market has exceeded $300 billion, nearly three times the full-year 2025 level. This year the US has seen nine deals each sized at least $25 billion, and apart from Skydance Corp.'s acquisition financing, all came from technology companies.

    #Industry news#English text is being prepared.

  17. CTech
    AI score78

    Anthropic Files Prospectus: 2025 Revenue of $4.59 Billion, Risk Section Warns AI May Threaten Human Survival

    Anthropic filed its prospectus ahead of its Nasdaq IPO, with 2025 revenue of $4.59 billion, up 1087.8% year over year, while operating losses widened 170.5% to $8.06 billion. In the first half of 2026, revenue was $16.29 billion, up 1300.3% year over year, and operating losses narrowed to $693 million. The prospectus devotes nearly 100 pages to listing risks, warning that AI without adequate safeguards could pose catastrophic or existential risks to humanity, while also stating that such risks could harm the company's business.

    Same story · 128 reports

    #Industry news#Safety / alignment#OpenAI

  18. 404 Media
    AI score60

    Proton will re-interview employees about their AI usage, and the results may affect whether they stay or go

    Proton told employees on Monday that it will conduct company-wide re-interviews on how employees use AI. Internal emails described these interviews as "having consequences," with evaluation results to be included in organizational and talent reviews that could lead to changes in roles, responsibilities, and employment relationships. The emails show that CEO Andy Yen has already interviewed more than 40 senior leaders in the engineering and product departments first; the re-interviews are scheduled for November 4 to 11, with interviewers drawn from a pool of leaders outside their teams and reporting lines who are deemed to have strong AI capabilities.

    #Industry news#Anthropic#English text is being prepared.

  19. TechSpot
    AI score93

    Anthropic Files S-1 for IPO, Plans Nasdaq Listing

    Anthropic has submitted its S-1 filing to the SEC, planning to list on Nasdaq under the ticker symbol ANT. The number of shares to be offered and the price range have not yet been disclosed; earlier reports said the target raise could be as high as $100 billion. The filing shows revenue of $16.3 billion in the first half of 2026, with second-quarter revenue of $11.6 billion, up 1368% year over year, while compute and infrastructure spending rose 300% over the same period. Shares held by the seven co-founders together carry 50.1% of shareholder voting power.

    Same story · 128 reports

    #Industry news#OpenAI

  20. The Decoder
    AI score83

    Anthropic Files S-1 Prospectus: First-Half Revenue of $16.3 Billion, Compute Commitments Exceed $780 Billion

    Anthropic has publicly filed its S-1 prospectus with the SEC, planning to list on Nasdaq under the ticker ANT, with Goldman Sachs and Morgan Stanley leading the underwriting; Yahoo Finance reports that Wall Street expects a valuation of about $2 trillion. Revenue grew from $376 million in Q1 2025 to $11.6 billion in Q2 2026, totaling $16.3 billion for the first half of 2026, and Q2 marked the first GAAP operating profit of $974 million. Compute contracts signed with Broadcom, Google, Amazon, Microsoft and others total approximately $780 billion to $810 billion; costs are largely fixed, while revenue fluctuates with usage.

    Same story · 128 reports

    #Industry news#OpenAI#Anthropic