AI Frontier

Also: OpenAI revenue tops $10B, CoreWeave shifts funding, and Meta loses infrastructure head

Nvidia's board could decide to remove founder and CEO Jensen Huang as soon as Wednesday evening. The board is reportedly preparing to oust him after he admitted to an affair with a woman accused of being a Chinese intelligence agent. Huang has led the company for 33 years. The scandal has already pushed Nvidia's stock down. This would be a major leadership shakeup at the world's most valuable AI chipmaker.

TSMC's board approved plans to invest one billion dollars in OpenAI and Anthropic. The Taiwanese chipmaker will put up to 500 million dollars into OpenAI preferred stock and another 500 million into Anthropic common shares at the expected IPO price. TSMC is the world's largest contract chipmaker and a key supplier for AI data centers. The move gives it direct stakes in two leading AI developers.

OpenAI's quarterly revenue reportedly crossed ten billion dollars for the first time. That is an almost 50 percent jump from the 6.7 billion dollars it recorded in the second quarter. The figure comes from a report, not an official announcement. It shows strong demand for the company's AI products and services.

CoreWeave is increasingly asking customers to pay up front and borrowing against specific contracts to fund its AI growth. The cloud provider is also looking at short-term deals that could lead to higher prices. This was disclosed in its third-quarter earnings. The shift reflects the high cost of building AI data centers.

Meta is losing its longtime head of infrastructure. Santosh Janardhan, who joined the company in 2009, is stepping down. He oversaw the massive computing systems behind Meta's AI and social media services. No replacement has been named.

Sources: techcrunch.com · arstechnica.com · techcrunch.com · techcrunch.com · techcrunch.com

This episode includes AI-generated content.

Also: Meta launches Muse Ember, Claude Code leaks secrets, and a DeepMind tool verifies Signal

Anthropic is moving forward with plans for an initial public offering. The AI company is expected to set its valuation range this week, according to the Wall Street Journal. This would make Anthropic one of the most valuable AI startups to go public. The move comes as demand for AI models and infrastructure remains strong.

Anthropic also released two new models called Mythos and Fable. The company says they show improvements in coding, reasoning, and vision. The models also communicate more clearly with users. This release comes amid concerns about recursive self-improvement in AI systems.

Meta released a new lightweight model for AI agents called Muse Ember. It costs ten cents per million input tokens. That makes it significantly cheaper than frontier competitors like Haiku and Luna. The model is designed for executing tasks on behalf of users.

Security researchers found that Claude Code is leaking sensitive information on public GitHub. The issue stems from how the tool handles user approval for automated shell commands. Thousands of secrets were exposed, including database passwords and API tokens. Anthropic has not yet commented on the findings.

Google DeepMind researchers used a new AI tool to formally verify the Signal encryption protocol. The tool proved that the widely used protocol works as intended. DeepMind says its stretch goal is to do the same for all software in the world. This could make software more secure by catching flaws before they are exploited.

Sources: techcrunch.com · techcrunch.com · techcrunch.com · arstechnica.com · wired.com

This episode includes AI-generated content.

Also: Trump signals openness to AI guardrails and world models gain attention

Nvidia is investigating CEO Jensen Huang over his relationship with a woman suspected of working for Chinese intelligence. Huang admitted in a memo to employees that he had an affair. The board will look into the matter. One expert puts the odds of Huang keeping his job at about fifty percent. This follows reports that Chinese spies targeted the chipmaker's chief executive. The investigation raises questions about leadership at a major AI company.

The New York Times added an AI-powered recipe modifier to its Cooking app. The beta feature helps users swap ingredients or adjust recipes to fit their needs. It is available now to some subscribers. The tool mostly works well, according to a review. This is a practical use of AI for everyday cooking.

President Trump said AI risks can be managed and signaled openness to guardrails. The statement came as governments debate how to regulate artificial intelligence. Trump did not propose specific rules. His comments suggest a shift from a hands-off approach.

World models are gaining attention as a way to teach robots and self-driving cars about the real world. These AI systems simulate physics in three dimensions. They could help machines navigate and interact with their environment. The technology is still developing but may become widely used.

Sources: techcrunch.com · wired.com · reuters.com · wired.com

This episode includes AI-generated content.

Also: Texas data centers expand and Democrats plan AI legislation

US intelligence circulated a memo claiming China's Ministry of State Security recruited a female technology executive to spy on Nvidia's CEO. The memo was reportedly shared earlier this year. It does not name the executive or the CEO, but Nvidia is a leading AI chipmaker. The claim highlights concerns about industrial espionage in the AI sector.

AI tools are increasingly appearing in apps for family caregivers. These apps help manage medications, appointments, and daily tasks. However, experts say the technology does not address caregivers' own well-being. The broader problem of caregiver burnout remains unsolved. The apps are becoming more capable, but their impact on stress is unclear.

Texas is seeing growth in AI infrastructure with new facilities in Dallas and College Station. The projects will add data center capacity in the state. This expansion reflects continued investment in AI computing power. Local officials have not disclosed the size or cost of the facilities.

Democrats in Congress are planning legislation for artificial intelligence. The proposal aims to set rules for AI development and use. Details of the bill have not been released. The move signals growing interest in federal AI regulation.

Sources: techcrunch.com · wired.com · nbcdfw.com · gvwire.com

This episode includes AI-generated content.

Also: Meta pauses AI training terms, Amazon lets users text Alexa, and a tiny speech model from Underdog

Crusoe is selling its troubled Wyoming data center project to Google. The deal highlights the challenges of developing massive AI data centers. Crusoe had faced difficulties completing the project. Google will take over the site, though terms were not disclosed. This transfer reflects the intense demand for data center capacity despite local hurdles.

OpenAI has reportedly raised an additional 11 billion dollars at a valuation of 1.25 trillion dollars. The funding round has not closed yet but is expected to in the coming weeks. This adds to OpenAI's already massive capital raise. The company continues to attract enormous investment for its AI models and infrastructure.

Meta has indefinitely delayed plans to update its commercial terms for business accounts. The proposed changes would have required extensive AI training on external websites linked on Facebook. Legal experts warned the language could allow Meta to ignore robots.txt scraping preferences. Meta said it is pausing the changes to engage with stakeholders. Existing terms remain in effect.

Amazon now lets users text Alexa to help them shop. The feature will roll out to millions of users in coming weeks. This follows Amazon's decision to block the Muse AI agent. Amazon is expanding Alexa's shopping capabilities through text interactions. The move aims to keep Alexa competitive in the AI assistant market.

Underdog has demonstrated a new speech model that is only 32 megabytes in size. Despite its small size, the model can generate both speech and music. This could make AI audio generation more accessible on devices with limited resources. Underdog has not said when the model will be released.

Sources: techcrunch.com · techcrunch.com · arstechnica.com · techcrunch.com · techcrunch.com

This episode includes AI-generated content.

Also: Meta adds video chat to Muse, OpenAI releases cybersecurity model, and Dropbox buys Tigris Data

Airbnb will launch an AI travel agent in the first half of 2027. The agent will help travelers plan trips, compare options, manage bookings, and get assistance during their stay. Airbnb announced the timeline but did not reveal details about how the agent will work. The move puts Airbnb in competition with other companies building AI-powered travel tools.

Agility Robotics began trading on the Nasdaq on November 5. It is the first pure-play humanoid robot company to go public in the United States. The company previously became the first to achieve full-time commercial deployment of humanoid robots at a customer warehouse in 2024. Its securities filing acknowledges that humanoid robots remain unproven at scale compared with industrial robots.

Meta now lets US users video chat with their Muse AI avatar. The AI agent can take calls as a customizable animated avatar. This comes ahead of the launch of a Tamagotchi-like wearable designed to bring Muse to life. Meta did not say when the wearable will be available. The feature expands Meta's push into personal AI companions.

OpenAI released a new cybersecurity model. The company did not provide details about the model's capabilities or how it differs from previous tools. The release comes as AI labs face scrutiny over their work to protect digital infrastructure. OpenAI has not said whether the model is available to the public or only to select partners.

Dropbox is acquiring Tigris Data for 350 million dollars. The deal will support the development of DBX Agent Cloud, a new storage platform for developers building AI models and agents. Dropbox expects the acquisition to bolster its AI infrastructure. The company did not say when the platform will launch.

Sources: techcrunch.com · arstechnica.com · techcrunch.com · techcrunch.com · techcrunch.com

This episode includes AI-generated content.

Also: Uber says almost nobody books rides with AI agents, and Meta acquires AI coding startup Conductor

Voters across the United States rejected data center expansion in Tuesday's midterm elections. Of 44 local measures across ten states seeking to restrict data centers, 38 passed. That includes dozens of moratoriums and bans. Analysts say the measures will mostly affect future projects planned for 2028, not those already under construction or approved through 2027. The results signal growing public opposition to the AI industry's rapid infrastructure buildout.

Apple is cutting its internal AI spending cap by one third. The company is significantly reducing the amount it spends on AI for its engineering teams. Apple has not explained the reason for the cut. The move suggests Apple is tightening budgets as it faces pressure on costs. It also raises questions about the company's AI development priorities.

Uber says almost no one is using AI agents to book rides. During its busiest week last quarter, only four people booked a ride using an AI agent. The company revealed the number to show that agent-based ride hailing is still extremely rare. This contrasts with the industry's push to make AI agents handle everyday tasks. Uber did not say whether it plans to invest more in the feature.

Meta has acquired AI coding startup Conductor. The startup lets software engineers deploy teams of AI agents to work on different parts of a project at the same time. Meta did not disclose the acquisition price. The deal adds agent-based coding tools to Meta's internal development. It reflects the race among tech giants to boost software productivity with AI.

Sources: arstechnica.com · pymnts.com · techcrunch.com · techcrunch.com

This episode includes AI-generated content.

Also: Fal doubles valuation to $16B, Netflix adds mood-based AI discovery, and Google ties school AI opt-out to losing Search

Anthropic is working on a system called FlashProve to prove an AI model's identity. The system would let anyone check that an AI response really came from the model it claims to be from. This could help combat impersonation and increase trust in AI outputs. Anthropic did not say when FlashProve will be available.

Data center builders are turning to old jet engines to generate power. The wait for grid connections and new turbines now stretches for years. These repurposed engines are less efficient and less environmentally friendly than modern turbines. But they can be deployed quickly to meet surging AI electricity demand. The trend highlights the severe power shortage facing the industry.

AI inference startup Fal has doubled its valuation to 16 billion dollars. The five-year-old company raised a 1.3 billion dollar round led by Bessemer. Fal provides infrastructure for running AI models in production. The huge valuation reflects surging demand for inference services as more companies deploy AI applications. The funding will help Fal expand its capacity.

Netflix is launching a new AI Discover tab to help users find something to watch based on their mood. The first experimental feature will offer recommendations based on what a user is currently in the mood for. Netflix did not specify how the mood detection works. The move aims to improve content discovery and keep subscribers engaged. The tab is rolling out experimentally.

Google will let schools disable its AI tools, but they will lose access to Google Search. The company is making AI tools opt-in by default for K-12 education products. However, because AI features are now integrated into Search, blocking AI means blocking Search entirely. Schools face a difficult choice between AI restrictions and search access. Google says the changes will roll out in the coming months.

Sources: techcrunch.com · techcrunch.com · techcrunch.com · techcrunch.com · techcrunch.com

This episode includes AI-generated content.

Also: Google expands Dreambeans in India, NYT sues Brave, and Musk throttles Grok Bot

Anthropic says it is very close to recursive self-improvement capability, also called RSI. RSI is the point where AI becomes good enough at training itself to trigger a dramatic acceleration in capabilities. The warning comes from the company behind the Claude models. Anthropic did not give a timeline for reaching RSI. The statement is likely to intensify debates about AI safety.

Apple is developing a home robot that could launch by 2027. The robot would roam around the house, support FaceTime calls, and switch on lights when a person enters a room. It may cost around 1,000 dollars or more. People familiar with the project told The Information that the robot is code-named J595. Apple's new CEO John Ternus is reportedly pushing the development.

Google is expanding its Dreambeans AI experiment to India. Dreambeans generates personal stories and now supports 13 Indian languages. The move aims to bring AI-generated storytelling to a wider audience. Google did not specify which languages are included. This is part of Google's effort to make its AI tools more accessible globally.

The New York Times and other outlets are suing the privacy-focused browser Brave. The lawsuit alleges Brave scraped copyrighted content and distributed it to customers like Anthropic and Perplexity. The publishers claim this violates their rights. Brave has not yet responded publicly. The case adds to a growing number of legal battles over AI training data.

Elon Musk is throttling limits on Grok Bot due to high costs. Grok Bot is the AI assistant from xAI. Musk said the product has seen the fastest growth he has ever witnessed. The move suggests that serving the bot is becoming expensive. It is unclear how much the limits will affect users.

Sources: techcrunch.com · arstechnica.com · techcrunch.com · techcrunch.com · techcrunch.com

This episode includes AI-generated content.

Also: HSBC opens Hong Kong AI institute and Microsoft offers buyouts to 60,000

OpenAI is considering rebranding its data centers as AI compute campuses. The move comes as data centers face public opposition over land use and energy demands. A new name could soften criticism and help win approval for future projects. OpenAI has not confirmed the plan. The change would be cosmetic, not technical.

South Korea will double its spending on AI and chips in 2027. The government announced the plan as part of a broader push to lead in advanced technology. Details on the exact amount were not provided. The increase signals strong state support for the semiconductor industry. South Korea is home to major chipmakers like Samsung and SK Hynix.

HSBC will open an AI institute in Hong Kong following a similar launch in Singapore. The bank aims to train employees and explore AI applications in finance. The institute will focus on practical skills and responsible use. HSBC did not specify a timeline for the opening. The move reflects growing investment in AI talent by global banks.

Microsoft is offering buyouts to 60,000 employees amid its AI investment push. The company is restructuring roles as it shifts resources toward AI development. The buyouts are voluntary and open to a wide range of positions. Microsoft has been investing heavily in AI infrastructure and partnerships. The move suggests a significant workforce adjustment.

Sources: techcrunch.com · japantimes.co.jp · mercedsunstar.com · mashable.com

This episode includes AI-generated content.