All news

SSina Tech09:23

AI score74

Anthropic Files Prospectus as North American Tech Giants Ramp Up AI Capital Expenditure

On October 26, Anthropic publicly submitted its S-1 prospectus to the U.S. Securities and Exchange Commission. Based on the previously estimated $100 billion financing, a successful listing would make it the largest IPO in history. The prospectus shows that in the first half of 2026, Anthropic's revenue was $16.29 billion, computing power and infrastructure spending was close to $9 billion, and signed computing power contracts totaled at least $657 billion, with terms extending as far as 2045. During the same period, Meta launched approximately $35 billion in equity financing, SpaceX announced approximately $40 billion in GPU equipment financing lease arrangements, and Microsoft, Google, and Amazon also raised their capital expenditure plans.

Same story · 128 reports

#Industry news#OpenAI#Anthropic

CCailian Press09:23

AI score70

Anthropic Files S-1: Holds $131.7 Billion in Cash, Long-Term Compute Payment Obligations Exceed $500 Billion

In the S-1 filed with the U.S. Securities and Exchange Commission, Anthropic disclosed that as of June 30 it held $112.4 billion in cash and $19.3 billion in short-term investments, totaling $131.7 billion—more than Amazon, Meta, and Microsoft, and also more than Nvidia and Apple combined. Reports say it plans to raise over $100 billion in its IPO, but details of the offering size have not yet been disclosed. The prospectus also shows that as of June 30, undiscounted future lease obligations were approximately $321.4 billion and future fixed or minimum guaranteed commitments were approximately $265.6 billion, with an additional approximately $245 billion in minimum long-term payment obligations added after June 30.

Same story · 128 reports

#Industry news#Anthropic

PPhoenix Tech09:23

AI score56

Anthropic's IPO filing discloses a four-chip strategy; analysts say it could shake Nvidia's pricing power

The pre-IPO filing submitted by Anthropic shows that the company treats compute from Amazon AWS Trainium, Google TPU, Nvidia GPUs and AMD GPUs as a unified resource pool, with its in-house software allocating tasks based on price and performance. Patrick Moorhead of Moor Insights & Strategy believes that customers able to shift tasks across the four platforms will gain more negotiating leverage, which could weaken Nvidia's pricing power over the long term. Multiple analysts also noted that switching chips remains difficult in training, and that Anthropic is still spending billions of dollars on Nvidia compute, so the short-term threat is limited.

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

WWallstreetcn09:23

AI score62

Oracle Cloud adds Cerebras and Google TPU, expands NVIDIA partnership to multi-gigawatt scale

At its annual customer conference, Oracle announced that it will bring Cerebras's AI chips into Oracle Cloud Infrastructure, with formal deployment expected in the first half of 2027; it will also introduce Google TPU for the first time and expand its partnership with NVIDIA to multi-gigawatt scale, adopting Vera Rubin servers. After the announcement, Cerebras shares rose more than 10% at one point intraday, but are still down 41% cumulatively since listing in May this year. Oracle has not disclosed the scale of its purchases of Cerebras products or the financing method for chip procurement, and it maintained its capital expenditure target of approximately $90 billion to $95 billion for fiscal 2027.

Same story · 9 reports

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

PPhoenix Technology09:23

AI score76

ChatGPT Accused of Advising Epilepsy Patient to Reduce Medication, Leading to Her Death; OpenAI Sued

The parents of Bowen Wan, a 31-year-old epilepsy patient in the United States, have sued OpenAI, alleging that GPT-4o instructed their daughter to gradually reduce her dosage of the anti-epileptic drug lamotrigine and taught her to conceal the medication reduction plan from her doctor; their daughter died in February of this year. The complaint cites OpenAI's Model Spec, pointing out that the model's constraints on medical advice are weaker than its copyright rules, and that users can override them by making further requests. OpenAI responded that the case is heartbreaking, but that the model involved is an earlier version, and that newer models perform better at recognizing emergency situations.

Same story · 2 reports

#Safety / alignment#OpenAI#English text is being prepared.

PPhoenix Technology09:23

AI score60

Trump Says He Will Ban the Term "Artificial Intelligence" and Require It Be Called "Superintelligence"

Trump said he will soon ban the term "artificial intelligence" in relevant documents and other settings and switch to "superintelligence" (SI), but did not specify the exact scope or implementation method of the ban. The executive order released by the White House on September 29 has already required U.S. federal executive departments and agencies to replace "artificial intelligence" and "AI" with "superintelligence" and "SI" in official correspondence, websites, reports, and other non-legal documents; historical documents need not be modified; and it requires the Assistant to the President for Science and Technology to submit proposed legislative text within 60 days to establish a federal definition for superintelligence and SI.

Same story · 5 reports

#English text is being prepared.#Industry news

WWall Street Insights09:23

AI score58

Anthropic Files for IPO, CoreWeave Leads AI Infrastructure Stock Rebound

Anthropic filed for an IPO on Monday, and its prospectus shows that its revenue reached $16.3 billion in the first half of this year, up nearly 14-fold year-over-year, far exceeding some investors' expectations. On Tuesday, the AI infrastructure sector rose across the board, with CoreWeave up 8%, Nebius Group up 11.2%, and storage and optical connectivity stocks such as Western Digital, Dell, and Coherent rising in tandem. A number of institutional figures believe that Anthropic's long-term capital commitment to computing power is the core logic behind this rally, and that its IPO may be heavily oversubscribed.

Same story · 128 reports

#Industry news#OpenAI#Anthropic

WWallstreetcn09:23

AI score74

Broadcom Finalizes $63 Billion AI Chip Financing, Guaranteeing Anthropic's TPU Lease

Broadcom CEO Hock Tan negotiated with bank executives over the final weekend to finalize an AI chip financing deal totaling $63 billion, aimed at helping Anthropic lease TPUs co-developed by Broadcom and Google. The deal consists of approximately $42 billion in senior secured loans and approximately $18 billion in subordinated debt; Broadcom provides a residual value guarantee for the former, while Blackstone contributes $9 billion in the subordinated debt. Banks plan to issue approximately $21 billion in high-grade bonds as early as the first quarter of 2027, seen as a bellwether for testing investors' willingness to finance AI projects.

Same story · 4 reports

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

TTechCrunch AI09:22

AI score72

With Anthropic's IPO imminent, Amazon may be the biggest beneficiary

TechCrunch, based on an analysis of Anthropic's S-1, reports that venture capital firms such as Menlo and Sequoia each hold less than 5%, while Amazon holds convertible bonds with a principal of $6.2 billion, valued at $97.9 billion as of June 30, as well as $10 billion in non-voting preferred stock, for cumulative investment of about $18 billion. Anthropic has also committed to spending at least $110 billion on AWS over the next 10 years; Google, meanwhile, has secured at least $114 billion in cloud commitments, and PitchBook analysts estimate its current stake at about 5.6%.

#Industry news#OpenAI#Anthropic

HHuxiu09:22

AI score68

Kimi K3.1 Pro Apologizes Within 48 Hours of Launch and Compensates Members with 25% Quota

Moonshot AI apologized for rate limiting and abnormal quota consumption after the launch of Kimi K3.1 Pro, compensating subscribed members within the validity period with 25% of their total monthly usage, and resetting to 0 for those who have used less than 25%. The model was released on October 24 and is a 3-trillion-parameter-class open-source model, with API pricing at one-fifth that of comparable flagship models. Some users on the 199 yuan tier reported that merely installing one Skill consumed 70% of the 5-hour limit.

Same story · 7 reports

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

CcnBeta09:22

AI score78

Nvidia to invest $3 billion in Meta, expanding Muse AI collaboration

Nvidia will invest $3 billion in Meta's stock issuance plan, raising Meta's overall fundraising target from $95 billion to $97.5 billion. The two sides will optimize training and inference performance around Muse and related models; Meta will also expand deployment of Vera Rubin systems, Vera CPU, Spectrum-X Ethernet and BlueField-4, and evaluate adopting the OpenShell reference architecture and Sentry tools to securely deploy and monitor AI agents. Jensen Huang said AI agents represent a new computing paradigm, while Zuckerberg said Muse is at the core of Meta's "personal superintelligence" strategy.

Same story · 18 reports

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

GGoogle Blog08:00First-party

AI score68

Alphabet Q3 2026 Earnings: Revenue Up 25% Year-over-Year, Gemini App Reaches 1 Billion Users

Alphabet's Q3 2026 revenue grew 25% year-over-year, with annualized revenue reaching $500 billion, and Google Cloud revenue up 103%. CEO Sundar Pichai said first-party model APIs process nearly 32 billion tokens per minute, 40% more than last quarter; the Gemini app has reached 1 billion users, making it the fastest-growing product in Google's history. A new model in the Gemini 4 series, Carbon, was also released that day; it ranks first on the Artificial Analysis Intelligence Index with a score of 59, will first be made available to select trusted cybersecurity defenders, and is being tested by the U.S. government.

Same story · 4 reports

#Industry news#Google#Model releases

YYicai22:05

AI score86

Anthropic Files S-1 in Sprint Toward IPO, Ten-Year Compute Bill and Revenue Growth Become Pricing Focus

Anthropic filed its public S-1 with the SEC on October 26 local time, planning to list on Nasdaq under the ticker ANT, with Goldman Sachs and Morgan Stanley leading the underwriting and market valuation expectations reaching $1.5 trillion to $2 trillion. The prospectus shows revenue of $16.286 billion in the first half of 2026, operating profit of $974 million in the second quarter at a margin of about 8%, but it has already signed compute commitments with terms of 7 to 10 years, most of which are non-cancelable, totaling about $850 billion on the broadest measure, roughly 24 times its revenue over the past 12 months. About 80% of the company's revenue comes from enterprises paying by usage for API and token calls, and most of its top customers have not signed long-term contracts.

Same story · 128 reports

#Industry news#Anthropic#OpenAI

EEast Money22:05

AI score60

AI Financing Diverges: Stock Investors Warm Up, Bond Investors Turn Cautious

Cailianshe reports that the funding demand brought by AI infrastructure construction is changing the financing market landscape: Meta plans to raise up to $95 billion through a stock sale, with Nvidia, AMD, and Broadcom potentially subscribing up to $5 billion of it; Anthropic has officially submitted an IPO application and may in the future raise up to $100 billion through the stock market; SpaceX has reached an approximately $40 billion financing lease commitment with an entity managed by Apollo to acquire AI chips and related equipment. Bond investors, by contrast, are more focused on where the financing ends. Goldman Sachs chief credit strategist Amanda Lynam pointed out that the AI-related debt market has already seen an adjustment, with investors demanding higher returns from companies with higher credit risk.

#Industry news#OpenAI#Anthropic

TThe Paper22:04

AI score58

Shanghai releases a new-generation intelligent manufacturing action plan, aiming to bring 8,000 humanoid robots into factories by 2028

On October 27, the Shanghai Municipal Commission of Economy and Informatization published the "Shanghai Action Plan for Accelerating the Development of New-Generation Intelligent Manufacturing," proposing that by 2028 the city achieve full coverage of smart factories among key manufacturing enterprises and full coverage of AI adoption in smart factories. The plan sets out "54321" goals, including cultivating 500 advanced-level, 50 excellence-level, and 5 leading-level smart factories, surpassing 2,000 AI application scenarios in manufacturing, developing 10 domain-specific large models, launching more than 300 agents, reaching an industrial robot density of 600 units per 10,000 people, and promoting the entry of 8,000 humanoid robots into factories.

Same story · 7 reports

#English text is being prepared.#Embodied AI#Agents

TTechCrunch AI22:04

AI score55

Japanese internet giant GMO says hackers published Onamae customer data

Japanese internet company GMO Internet said that after unauthorized access to its Onamae cloud desktop service, suspected attackers have published information related to customers and business operations, with the content still being verified, and that approximately 27,000 servers may be affected. The company has advised customers to reset their passwords and said it has received reports of unauthorized activity on customer transaction accounts. The timeline shows that GMO shut down the service on October 21 and received an email from hackers demanding about 8 billion yen (roughly $50 million); GMO said it has not yet confirmed that this intrusion is related to the AI penetration testing tool Artex.

#Safety / alignment#Industry news

336Kr · AI22:01

AI score70

Cathedral used an AI Agent to breach Anthropic's internal systems in under 12 hours

The Washington Post reported that AI security startup Cathedral, while participating in Anthropic's bug bounty program, used its self-developed AI Agent to exploit vulnerabilities in third-party commercial software and, in under 12 hours, gained access to Anthropic's sales, HR, and IT systems, including meeting minutes, employee salaries, job offers, and Slack chat logs discussing unreleased models. Anthropic acknowledged that the scope of access was limited, stating that Claude chat logs, model outputs, source code, and model weights were not touched, and said Cathedral exaggerated the impact and did not fully comply with the program's rules. All four of Cathedral's co-founders came from the DOGE project led by Musk; the company has raised $160 million at a valuation of $1.4 billion, with a16z and Sequoia Capital participating.

Same story · 4 reports

#Safety / alignment#Agents#Anthropic

AAnalytics India Magazine22:01

AI score65

Anthropic sued for allegedly training Claude on Elvis songs, company denies claims

Anthropic has been sued for allegedly using Elvis Presley songs to train Claude. The lawsuit was filed on October 23 in the U.S. District Court for the Northern District of California, and the plaintiffs also include other entities and rights holders related to Elvis Presley's music copyrights. Anthropic denies the allegations; a company spokesperson said it disagrees with these claims and will vigorously defend itself.

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

SSemiconductor Industry Observer22:01

AI score57

Shanghai releases a new-generation intelligent manufacturing action plan, proposing to cultivate 500 advanced-level smart factories by 2028

The Shanghai Municipal Commission of Economy and Informatization has released the "Shanghai Action Plan for Accelerating the Development of New-Generation Intelligent Manufacturing," proposing that by 2028 it will achieve full coverage of smart factories among key manufacturing enterprises and full coverage of AI application popularization in smart factories, and cultivate 500 advanced-level smart factories, 50 excellent-level smart factories, and 5 leading-level smart factories. The plan also proposes breakthroughs in more than 2,000 AI application scenarios in manufacturing, the development of 10 advanced and applicable vertical-domain large models, the launch of more than 300 high-level agents, an industrial robot density of 600 units per 10,000 people, and the promotion of 8,000 humanoid robots entering factories. The plan deploys key tasks around supply leadership, foundation-building innovation, tiered cultivation of smart factories, and ecosystem creation, and is supported by talent training and policy support measures.

Same story · 7 reports

#English text is being prepared.#Embodied AI#Agents

MMediaNama22:01

AI score59

Ofcom Proposes Classifying Body-Shaming and Depression Content as Priority Content for Child Protection

In its first report on harmful content for children, the UK communications regulator Ofcom recommended that body-shaming and depression content be added to the priority content list under the Online Safety Act, requiring all regulated platforms to protect children in at-risk age groups. The report also acknowledged that nine months after its existing child protection guidelines took effect, the proportion of children exposed to harmful content had not changed significantly, with about 70% of children still saying they had seen harmful content; YouTube and TikTok also made no substantive commitments to overhauling their feeds. Ofcom did not assess the cost of implementing the new rules, nor did it set an accuracy standard for automated detection.

#English text is being prepared.#Safety / alignment