Hi there, this is your daily ☕️ Techpresso.
In today's Techpresso:
👓 Meta unveils smart glasses with display and neural wristband
🤝 Nvidia to invest $5B in Intel and develop chips with onetime rival
🥇 OpenAI, Google models take gold at ICPC contest
👀 Reddit wants a better AI deal with Google
⏸️ China suspends its antitrust probe into Google
🎁 + 17 other news you might like
🔮 + 6 handpicked research papers and tools
👓 Meta unveils smart glasses with display and neural wristband LINK
Meta announced its $799 Meta Ray-Ban Display smart glasses, which have a built-in screen on the right lens for showing directions, social media apps, and live translations.
The device is controlled by the Meta Neural Band, a screenless wristband that uses electromyography to read subtle hand gestures and let people navigate through different applications.
This consumer product is less capable than the company's Orion prototype, shipping without augmented reality lenses or eye tracking and using a much simpler display for alerts.
🤝 Nvidia to invest $5B in Intel and develop chips with onetime rival LINK
Nvidia plans to purchase a $5 billion stake in its former competitor Intel, kicking off a broad partnership to together develop new data center and consumer products.
The deal will produce custom x86 CPUs for Nvidia's AI platforms and new "x86 RTX SoCs" for PCs, which integrate chiplets of Nvidia's RTX GPUs into Intel's hardware.
Both companies will connect their distinct architectures using Nvidia's high-speed NVLink interface, a crucial link that enables faster data and control code transfers between processors.
🥇 OpenAI, Google models take gold at ICPC contest LINK
At the ICPC World Finals, Google’s Gemini 2.5 Deep Think earned a gold-medal score by solving 10 of 12 problems, even cracking one that stumped all human teams.
In a stunning result, OpenAI’s GPT-5 model reportedly achieved a perfect score by solving all 12 programming challenges, submitting the correct answer on its first attempt for 11 of them.
This performance shows the models have moved past just generating code, using creative, multi-step reasoning to solve complex algorithmic problems that once required human intellect to crack.
👀 Reddit wants a better AI deal with Google LINK
Reddit is renegotiating its AI partnership to replace its current flat-fee arrangement with Google with a “dynamic pricing” model that pays based on how its user-generated content powers AI answers.
The platform believes its archive of human conversations is undervalued and is using a report naming it the most-cited domain by AI models as leverage to get more compensation from partners.
This new structure acts as a hedge against the “AI Paradox,” where Google’s AI Overviews summarize answers and reduce the click-through traffic that is necessary for Reddit's advertising business to work.
⏸️ China suspends its antitrust probe into Google LINK
China’s State Administration for Market Regulation is shelving its competition inquiry into Google amid delicate trade negotiations between Washington and Beijing over both TikTok and the chipmaker Nvidia.
This probe will now enter a status in Chinese called “zhongzhi,” although the technology company has not yet received any formal paperwork from Beijing confirming the case's closure.
The action coincides with China's internet watchdog directing major tech firms to halt tests and purchases of Nvidia’s RTX Pro 6000D, a product designed for the local market.
Other news you might like
- Meta Connect 2025: the 6 biggest announcementsLINK
- DeepSeek reveals its popular AI R1 model cost just $294,000 to train, challenging US AI giants on price and chipsLINK
- Google revamps Discover page to show content from creatorsLINK
- Google will use hashes to find and remove nonconsensual intimate imagery from SearchLINK
- US-banned Huawei unveils ‘world’s most powerful’ AI cluster to boost China techLINK
- Tesla is redesigning its door handles following safety probe, Bloomberg investigationLINK
- Nvidia AI chip challenger Groq raises even more than expected, hits $6.9B valuationLINK
- Zoom launches a cross application AI notetaker, AI avatars and more in its latest updateLINK
- Valve, Discord, Reddit, and Twitch CEOs called to testify before Congress on online radicalizationLINK
- Nothing wants you to talk to your earbuds’ charging caseLINK
- 'Ask Gemini' AI will tell you what you missed during a Google Meet callLINK
- Samsung brings ads to US fridgesLINK
- Inside the Apple audio lab where AirPods are tested and tunedLINK
- Genetic bioengineering firm steps closer to reviving the dodoLINK
- Flying cars crash into each other mid-air in ChinaLINK
- Amazon is creating more tools to fill its site with AI adsLINK
- AI models know when they're being tested - and change their behavior, research showsLINK
Latest research and tools
Towards a Physics Foundation Model: the paper proposes creating a large-scale AI trained on diverse physics data to serve as a general-purpose tool for scientific tasks.LINK
Gluon: a GPU programming language based on the same compiler stack as Triton.LINK
Milwaukee M18 tool battery communication protocol: a tool that communicates with Milwaukee M18 batteries to let users read diagnostic information and generate a health report for the battery on their computer.LINK
Jqp: a terminal-based playground for interactively writing and testing jq queries.LINK
Automating Influence: The Emergence of AI Propaganda Factories: large language models can be used to create systems that automatically produce vast amounts of persuasive propaganda.LINK
SpikingBrain Technical Spiking Brain-Inspired Large Models: this paper introduces a large language model that uses brain-inspired spiking neurons to operate with high energy efficiency.LINK
Want to get the latest news differently? Find us on:
See you tomorrow for a new dose of ☕️ Techpresso!
.png){kind=small}
.png){kind=small}
.png){kind=small}