AI Daily Newsletter 18/07/2026: Meta rents GPUs to rival Anthropic and Linus Torvalds fires back at anti-AI group

🤝 Zuckerberg Allows Anthropic to Rent Excess GPU Computing Power

Meta is in talks with Anthropic to rent out excess GPU computing power from its massive data centers. According to sources, the deal is estimated to be worth up to $10 billion over two years. Anthropic is said to have proposed the idea in June 2026, and while Meta is carefully considering it, the agreement could still fall through if either party decides to back out.

For Meta, this is an opportunity to open up a new massive revenue stream to offset its huge investment in equipment. CEO Mark Zuckerberg has shared with investors that Meta is willing to sell or rent out excess computing power if the demand for AI in its internal applications does not grow fast enough to keep up with the pace of hardware purchases. This year alone, Meta is expected to spend $145 billion on AI infrastructure. Renting out GPUs to rival Anthropic allows Meta to quickly monetize its expensive H100 chips, while Anthropic can immediately access resources to train its next-generation Claude model without having to rely on or wait for traditional cloud providers.

Source: The Decoder

💻 Linus Torvalds Tells AI Critics in the Linux Community to “Fork Off”

Linus Torvalds, the creator of the open-source Linux operating system, has made some scathing comments in defense of using AI tools in the Linux kernel development process. On the Linux kernel mailing list, Torvalds stated bluntly: “Linux is not an anti-AI project.” He emphasized that he will use his voice as the top maintainer to support useful tools and sent a message to AI critics that they can fork the project or leave.

The debate erupted over Sashiko, an AI-based automatic code review tool for the Linux kernel, sponsored by the Linux Foundation. When some developers expressed strong opposition, citing that AI is not reliable and could damage the project’s reputation, Torvalds immediately spoke out. He acknowledged that current AI can still cause certain annoyances, such as creating extra work for maintainers or pointing out silly programming errors, but rejecting it is like “sticking your head in the sand and screaming loudly so you can’t hear anything.” According to Linus, anyone who doubts the value of AI simply hasn’t used it in their work.

Source: The Decoder

🛡️ Patreon Stops Asking AI Bots Not to Scrape and Starts Blocking Them

Patreon, a platform that supports creators, has officially announced a partnership with Cloudflare to actively block AI data scraping bots that scan creators’ artwork and content without permission to train models. This move marks a significant shift for Patreon from using traditional robots.txt files to actively preventing copyright infringement with strong technical measures.

According to Patreon’s representative, upgrading their shield is essential as AI data scraping techniques become more sophisticated and intentionally ignore ethical guidelines. During the trial of Cloudflare’s AI Crawl Control feature, the number of attempts by AI training bots to access Patreon’s system dropped from thousands per week to zero. Drew Rowny, Patreon’s Product Director, said that most of the internet environment forces creators to accept AI data scraping as a price to pay for reaching fans. Patreon aims for a different future where creators have the right to speak up and control how their work is shared and used.

Source: TechCrunch

🔇 A Clever Trick to Say No to AI Transcription and Recording on Zoom

In the context of AI-powered meeting transcription and summary apps booming, Jeremy Levine, a managing partner at Bessemer Venture Partners, has found a unique method to protect his privacy. When attending online meetings, he no longer uses his real name simply but changes it to: “Jeremy Levine I do not consent to transcribing or recording.”

This trick reflects growing concerns in the tech community over the wave of automatic transcription and summarization of every aspect of life, from work to personal. Some investors said they have to assume that all their meetings are being recorded without prior notice. Even a founder revealed that she used the AI app Granola to record her first dates and asked Claude to analyze whether she was empathetic and charming enough. However, Levine considers this widespread recording behavior as “socially unacceptable” and completely destroys the open and spontaneous atmosphere of discussions. It also creates a massive audio data dump when dozens of recording files are stored but nobody has time to listen to them, accompanied by complex legal risks regarding privacy.

Source: TechCrunch

🤖 Agility Robotics Opens Robot Training Center Next to Tesla

Agility Robotics has decided to expand by establishing a 5,500-square-meter robot training center in Fremont, California. Notably, the location is right next to Tesla’s manufacturing plant, where Elon Musk is expected to start mass-producing the Optimus robot this year. This is seen as a direct challenge that Agility sends to Tesla right in the electric car giant’s backyard.

Although Agility does not have the vast financial resources like Musk’s company, they are confident in leading due to their Digit robot already being commercialized and generating a stable revenue stream. Digit is currently unloading cargo at logistics facilities of Amazon, GXO, Schaeffler, and Toyota, bringing in orders worth over $300 million for the company. CEO Peggy Johnson shared that locating the facility in Fremont helps the company attract high-quality talent in the automation industry. Notably, Agility insists on not using Generative AI for their core safety control system to avoid robots “inventing” unexpectedly when working alongside humans, although Generative AI is still applied at a higher level to program complex tasks on a large scale.

Source: TechCrunch