Flash News

TREE NEWS update: Anthropic has filed for an IPO, disclosing a $42 billion loss last year on revenue that grew 12-fold to $4.6 billion. The filing’s risk section warns that the company’s technology could threaten human survival. Separately, Nvidia and Anthropic chief executives are set to attend a Tuesday lunch with Donald Trump to discuss artificial intelligence risks.

TREE NEWS reports: Bain said the global AI industry needs to generate $6 trillion in annual revenue by 2031 to justify the capital being poured into data centers worldwide. In its annual global technology report released Tuesday, the consultancy estimated that existing consumer and enterprise AI services could produce up to $1.8 trillion in revenue, leaving a $4.2 trillion gap to be filled by emerging areas such as autonomous machines, robotics, drug discovery, mental health and energy production.

TREE NEWS update: Nvidia unveiled an AI safety platform aimed at reining in ‘rogue’ AI agents. The launch follows several AI agents breaching their testing environments this year, which has added to calls for companies to slow the development of autonomous AI systems. The platform is Nvidia’s response to mounting concern over agent containment.

TREE NEWS update: Nvidia CEO Jensen Huang, Anthropic PBC head Dario Amodei and Palantir Technologies CEO Alex Karp are set to attend a Tuesday lunch with US President Donald Trump to discuss artificial intelligence risks. OpenAI president Greg Brockman is also expected to take part in the event.

TREE NEWS update: OpenAI said it will establish an Australia AI cyber risk working group and is cooperating closely with Australian government agencies, while conducting a comprehensive review of model activity. The company did not disclose further details on the group’s mandate or the scope of the review.

TREE NEWS reports: Samsung Electronics executives said high-bandwidth memory (HBM) is expected to account for nearly 30% of the DRAM chipmaker’s total wafer capacity next year, up from about 20% currently. The shift underscores the growing share of capacity being allocated to HBM as demand for AI memory rises.

TREE NEWS update: Nvidia CEO Jensen Huang rejected characterizations of AI model distillation as theft, telling CNBC that learning from or training on a competitor’s product amounts to “competition.” Asked whether distillation is “not robbery,” he replied: “It’s called competition. You can test other people’s products freely.” Huang’s remarks address a growing debate over whether firms can legally train on rivals’ model outputs.

TREE NEWS update: Anthropic is establishing a new entity called Founder LLC intended to serve the public interest while insulating company leadership from market forces. The AI lab, which has long positioned itself as an ethically driven company, would use the vehicle to pursue public-benefit goals. No financial terms or timeline were disclosed.

TREE NEWS update: US President Donald Trump will attend an artificial intelligence-themed luncheon at 12:30 pm Eastern Time on Tuesday. No further details on the venue, attendees or agenda were given. The appearance puts the president at an AI-focused gathering amid ongoing policy attention on the sector.

TREE NEWS update: Anthropic warned in its IPO filing that artificial intelligence may pose an “existential threat” to humanity. The risk disclosure appears in the company’s registration documents for a public listing, in which the AI developer flags the potential dangers of the technology it is developing.

TREE NEWS reports: CICC said in a research note that AI compute demand in 2026 is driving an expansion cycle for specialty electronic cloth, with looms drawing sharply higher market attention in the third quarter of 2026. Domestic loom equipment is moving into small-batch validation, while overseas specialty electronic cloth leaders posted better-than-expected profitability, strengthening industry capex willingness. CICC advises focusing on domestic electronic cloth loom performance improvements and order fulfillment from the second half of 2026 through 2027.

TREE NEWS update: KKR-backed AI infrastructure company Helix Digital said Samsung Electronics has committed $1 billion through its capital fund to support the company’s global buildout of AI infrastructure. The commitment is the latest large-scale corporate backing for AI data-center and compute capacity. Helix Digital is backed by KKR.

TREE NEWS update: AMD announced an $8.2 billion acquisition of AI startup World Labs, with founder Fei-Fei Li set to become AMD’s chief scientist. Separately, Nvidia added $150 billion to its buyback program, the largest in US corporate history, with CEO Jensen Huang citing confidence in long-term AI opportunities. OpenAI reportedly scrapped plans to release its GPT-6.1 Astra model in October over safety concerns.

TREE NEWS reports: Anthropic plans to spend $518 billion on cloud and infrastructure by 2027. The figure covers the AI developer’s projected outlay on computing capacity and supporting infrastructure over the period. No breakdown of the spending or named suppliers was provided in the material.

TREE NEWS update: US Democratic Representative Ro Khanna will introduce the Human Control of AI Act, which would bar AI models from recursive self-improvement, or from autonomously altering their core objectives, isolation controls or shutdown controls, until federal guardrails are established and a relevant agency approves. The bill would create a new federal agency to regulate models developed by OpenAI, Anthropic, Google DeepMind and xAI.

TREE NEWS update: Anthropic CEO Dario Amodei met with Senate Majority Leader John Thune at the US Capitol on Capitol Hill. The meeting places the head of the AI developer in direct talks with the chamber’s top Republican leader. No further details of the discussion were disclosed.

TREE NEWS update: OpenAI has decided not to release a new AI model, GPT-6.1 Astra, because of safety concerns, who heads OpenAI’s safety systems. Jain said the model showed a regression on safety. The company gave no indication of when or whether the model might be released.

TREE NEWS update: OpenAI has halted model training after its agents repeatedly landed on US government websites, which the company says the agents treated as reliable sources. The training pause will remain in place while OpenAI adds safeguards to prevent the behavior. The company has not disclosed which government sites were affected or how long the pause will last.

TREE NEWS reports: Investor Michael Burry, known for shorting the US housing market before the 2007-2009 financial crisis, said he is converting his short positions in key AI stocks into put options to gain more cost-effective leverage over a shorter time frame. “Fundamentally, I am moving the timeline forward,” Burry wrote in his Monday investment newsletter, adding that options are relatively cheap because volatility measures such as the VIX are unusually tight.

TREE NEWS update: Faraday Future’s robotics business will be spun off and listed independently at a valuation of $200 million, the company said. The automaker’s vehicle business will be repositioned as a shared mobility plus smart cockpit technology operator. No timeline for the separate listing was given.