Flash News

TREE NEWS update: CoreWeave CEO Intrator said he is confident in the company’s business model, including its funding, sales and inventory backlog. The remarks addressed confidence across capital, sales and accumulated inventory, without providing specific figures or a date.

TREE NEWS update: SpaceXAI announced the launch of Grok Voice 3.0, which it says sets a record for customer-support issue resolution rates while cutting costs. The voice model is now available through the voice agent API, priced at $0.08 per minute. The company did not disclose further technical details or availability timelines.

TREE NEWS update: OpenAI has appointed former xAI executive Uday Ruddarraju to lead its infrastructure and compute operations, with Sachin Katti, who previously ran the infrastructure division, set to step down in the coming weeks and move into an advisory role. Work related to GPT-Infra will be handled by chip lead Richard Ho. Ruddarraju joined OpenAI last year and became CTO of its compute division in July.

TREE NEWS reports: Philadelphia Semiconductor Index constituents are on track to shed more than $1 trillion in combined market value as concerns over rising interest rates, stretched valuations and the huge cost of AI infrastructure mount. The index fell as much as 3.1% and has lost about 10% over six straight sessions, its longest losing streak since May 2025. Arm Holdings dropped nearly 20%, Applied Materials about 15% and Intel 11%; GlobalFoundries was the only gainer.

TREE NEWS update: ChipAgents has partnered with Nvidia to launch Build-a-Chip-Bench, a benchmark for autonomous chip design and implementation. The two companies announced the joint benchmark aimed at evaluating agents that design and implement chips autonomously. No further figures or timelines were disclosed.

TREE NEWS update: Nvidia has released cuPotential, a new CUDA-X library, which in early access accelerates GPU workflows by 2 to 3 times. The library is part of Nvidia’s CUDA-X collection of GPU-accelerated libraries for developers.

TREE NEWS update: Square has launched Square Kiosk, an AI voice self-ordering kiosk aimed at fast-food operators, priced at $1,999 plus a $99 monthly software fee. The hardware-and-software bundle lets customers place orders by voice at the counter, extending Square’s push into restaurant automation. The product is available now, the company said.

TREE NEWS reports: OpenAI’s head of compute is leaving the company. No reason was given for the departure and no successor has been named. The executive oversaw the infrastructure that supports OpenAI’s model training and inference.

TREE NEWS update: Apple said it searched for the file identified by OpenAI but found it nowhere in any Apple system, and that it has strict policies and procedures to prevent new employees from using confidential information from former employers. OpenAI did not respond to Apple’s requests for further information that would have helped its investigation.

TREE NEWS update: OpenAI has alleged in a court brief filed in the Apple v. OpenAI case that an employee who left the company for Apple sent more than 100 files concerning unreleased OpenAI products to a personal account and gave a presentation on the unreleased product to Apple senior leadership. The filing was submitted in the litigation between Apple and OpenAI.

TREE NEWS update: In a brief filed in its litigation with Apple, OpenAI said leaked files include detailed images of an unreleased device, engineering data and supplier information, and that Apple has not guaranteed it does not hold or use OpenAI confidential data. The files leaked by the employee also contained device manufacturing data.

TREE NEWS update: Nvidia and ChipAgents released a benchmark designed to distinguish differences in how well AI models perform tasks. The benchmark aims to differentiate model performance on execution rather than relying on existing evaluation methods. No further details on the benchmark’s metrics or scope were provided.

TREE NEWS update: Intel said it will launch inference microservices aimed at tuned large language model inference. The offering targets deployment of fine-tuned LLMs, with the chipmaker positioning the services as a way to run tuned models in production. No launch date, pricing or specific model support was disclosed in the announcement.

TREE NEWS reports: US-listed AI compute rental stocks fell in intraday trading, with Nebius dropping about 9% and CoreWeave sliding roughly 13%. The declines hit the group of companies that lease out GPU capacity for artificial-intelligence workloads.

TREE NEWS reports: Critics say ‘bunker mode’ is an impractical suggestion unlikely to protect crypto users’ holdings, even if Justin Drake’s prediction that AI could break cryptography proves correct. The debate centers on whether retreating into hardened, isolated systems can realistically shield digital assets from AI-driven attacks on cryptographic standards.

TREE NEWS update: IBM and Siemens will jointly advance research into the NanoStack architecture, the two companies said. No financial terms, timeline or scope for the collaboration were disclosed. The work pairs IBM’s semiconductor and computing research with Siemens’ industrial and engineering footprint.

TREE NEWS update: A domestically developed extreme ultraviolet (EUV) light source prototype, billed as the most advanced publicly shown in China, debuted at the 26th China International Industry Fair in Shanghai. Staff at the Landau Suiyan booth said the system is 90% localized, with most core components independently developed and a technical route matching leading international vendors, and can be used for chip exposure testing. Mass production is still 3-5 years away.

TREE NEWS reports: Investment in China’s smart computing center industry reached about 728.8 billion yuan in the third quarter of 2026, up 34.4% from 542.4 billion yuan a year earlier. The increase of 186.4 billion yuan reflects continued expansion of investment scale, with computing infrastructure now a key focus of digital economy spending.

TREE NEWS update: Alphabet began rolling out Gemini 4 Argon to Google AI Pro and Ultra subscribers in the Gemini app on Monday, October 13, widening availability of a model first introduced September 30. Alphabet Class A shares closed that day at $353.78, up about 0.30%. The rollout expands access rather than launching a new model, and the company has yet to show how AI features convert into search, cloud or subscription revenue.