Hi there, this is your daily ☕️ Techpresso.
In today's Techpresso:
⚙️ OpenAI and Jony Ive's AI device delayed over technical issues
🤝 OpenAI signs massive GPU deal with AMD
🛡️ Google DeepMind unveils CodeMender, an AI agent that autonomously patches software vulnerabilities
🔄 Apple prepares for Tim Cook's retirement
🚗 Tesla hints at a new affordable car model
💸 Musk bets billions in Memphis to accelerate his AI ambitions
🎁 + 10 other news you might like
🔮 + 3 handpicked research papers and tools
⚙️ OpenAI and Jony Ive's AI device delayed over technical issues LINK
OpenAI and Jony Ive are creating a palm-sized, screen-less AI gadget that faces delays from unresolved issues with its personality, privacy, and required computing infrastructure.
The team is struggling with an “always on” approach because the hardware has difficulty knowing when to speak up and how to end conversations at the appropriate moment.
The product is designed to interpret audio and visual cues from the physical environment, responding to user requests without needing a display or any specific verbal prompts.
🤝 OpenAI signs massive GPU deal with AMD LINK
OpenAI has committed to purchasing 6 gigawatts of AMD processors, starting with the MI450 model next year, to power its AI data centers and cope with skyrocketing inference demand.
The partnership is expected to generate tens of billions in revenue for AMD, while OpenAI will receive warrants for up to 160 million company shares if it meets deployment milestones.
The first 1 gigawatt deployment of the MI450 chips is scheduled for the second half of 2026, with both companies calling the multi-billion-dollar arrangement "definitive" and ready for regulators.
🛡️ Google DeepMind unveils CodeMender, an AI agent that autonomously patches software vulnerabilities LINK
Google DeepMind's new AI agent, CodeMender, autonomously detects and patches software vulnerabilities by using Gemini Deep Think models combined with program analysis to rewrite unsafe code.
The research tool has already submitted 72 security fixes to open-source projects and applied “-fbounds-safety” annotations to the libwebp library to prevent buffer overflow exploits.
Its system uses an "LLM judge" to validate proposed changes, but every single patch is still reviewed by human researchers before being submitted to a project for approval.
🔄 Apple prepares for Tim Cook's retirement LINK
With CEO Tim Cook turning 65 soon, Apple's board is reportedly considering Senior Vice President of Hardware Engineering John Ternus as the leading candidate for his replacement.
The board might favor a leader with a technical background since Apple has stumbled in areas including generative AI, mixed reality, the smart home, and autonomous vehicles.
At 50 years old, Ternus is younger than other senior executives and is now getting more public exposure from the company, such as when he introduced the iPhone Air.
🚗 Tesla hints at a new affordable car model LINK
Tesla's mysterious teaser video has sparked speculation that the company will announce its new affordable model, which has been seen uncovered recently and is described as a stripped down Model Y.
The potential launch of a more affordable EV comes as the $7,500 tax credit ends, providing a logical way for the automaker to maintain growth in its upcoming delivery numbers.
Other theories suggest the October 7 announcement could instead be a demo for the delayed Roadster, which may feature a SpaceX package with cold-gas thrusters for rapid acceleration.
💸 Musk bets billions in Memphis to accelerate his AI ambitions LINK
Elon Musk’s xAI is spending billions to build the 'Colossus' supercomputer in Memphis, aiming to house up to 1 million Nvidia GPUs in a massive race against AI rivals.
The project faces a federal lawsuit from the NAACP over alleged illegal air pollution from natural-gas turbines built to power the site in a predominantly Black, low-income community.
Alongside the legal battle, xAI is experiencing internal turmoil with an "executive exodus" and product issues with its Grok chatbot, adding risk to its high-stakes AI gamble.
Other news you might like
- Deloitte to refund government over AI errorsLINK
- Apple faces cybercrime investigation in France after Siri complaintLINK
- Japanese tech giant deploys laser drones to protect chickens — drones are hoped to prevent the spread of avian fluLINK
- Hollywood is Already Falling Behind on AI VideoLINK
- Qualcomm in the dock over 'patent tax' on smartphonesLINK
- Rabbit plots its redemption arcLINK
- Rivian says ‘customers will appreciate’ lack of CarPlay eventuallyLINK
- OpenAI’s “Hacktivate AI” report urges Europe to cut red tape and harmonize digital regulationsLINK
- When people create Sora deepfakes of you, you can now set limitsLINK
- Elon Musk tries to make Apple and mobile carriers regret choosing Starlink rivalsLINK
Latest research and tools
Toybox: a project that combines many common Linux command line tools into one simple executable.LINK
Kent Dybvig's Scheme Machine: a compact implementation of the Scheme programming language written in only 400 lines of C code, demonstrating a minimalist design for running Scheme programs.LINK
Mise: a tool for running tasks across multiple projects contained within a single large codebase, also known as a monorepo.LINK
Want to get the latest news differently? Find us on:
See you tomorrow for a new dose of ☕️ Techpresso!