ร— โˆ’ + newsletter-2026-02-25
โ•”โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•—
โ•‘                                                             โ•‘
โ•‘   โ–€โ–ˆโ–€ โ–„โ–€โ–ˆ โ–ˆโ–„โ–€ โ–ˆโ–€โ–€   โ–ˆ โ–€โ–ˆโ–€   โ–ˆโ–€โ–€ โ–ˆโ–€โ–ˆ โ–ˆโ–€โ–ˆ โ–ˆโ–€โ–„โ–€โ–ˆ             โ•‘
โ•‘    โ–ˆ  โ–ˆโ–€โ–ˆ โ–ˆ โ–ˆ โ–ˆโ–ˆโ–„   โ–ˆ  โ–ˆ    โ–ˆโ–€  โ–ˆโ–€โ–„ โ–ˆโ–„โ–ˆ โ–ˆ โ–€ โ–ˆ             โ•‘
โ•‘                                                             โ•‘
โ•‘   โ–€โ–ˆโ–€ โ–ˆ โ–ˆ โ–ˆโ–€โ–€   โ–€โ–ˆโ–€ โ–ˆโ–€โ–ˆ โ–ˆโ–€โ–ˆ                                โ•‘
โ•‘    โ–ˆ  โ–ˆโ–€โ–ˆ โ–ˆโ–ˆโ–„    โ–ˆ  โ–ˆโ–„โ–ˆ โ–ˆโ–€โ–€                                โ•‘
โ•‘                                                             โ•‘
โ•šโ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•
    

// Tech news at terminal velocity

cat 2026-02-25.md

Billions, Bots, and Boozy Chimps

$ cat TLDR.md

โ–ธ Wayve secures a massive $1.2B to teach cars how to drive

โ–ธ MatX grabs $500M to take a swing at the Nvidia empire

โ–ธ Cloudflare rewrites Next.js on Vite in a week because they can

Headlines & Launches

๐Ÿš— Wayve raises $1.2B to make AVs smarter, not just mapped

The UK self-driving startup just backed up the Brinks truck with help from Nvidia and Uber. Unlike competitors relying on HD maps, Wayve is betting on 'embodied AI' that learns to drive like a humanโ€”hopefully better than the one cutting you off in traffic.

Source: TechCrunch

๐Ÿ’ฐ MatX raises $500M to challenge the Green Team

Founded by ex-Google TPU engineers, MatX is building chips specifically for LLMs, claiming they can do it better and cheaper than Nvidia. A half-billion-dollar seed round suggests investors are desperate for a Plan B in the silicon wars.

Source: TechCrunch

๐Ÿ“ Honor's MagicPad 4 is thinner than your patience

At just 4.8mm thick, the new MagicPad 4 is the world's thinnest Android tablet, beating its predecessor by a full millimeter. Itโ€™s an impressive feat of engineering, assuming you don't accidentally snap it in half while putting it in your bag.

Source: The Verge

โš–๏ธ OpenAI wins a round against Elon's xAI

A judge dismissed xAI's trade secrets lawsuit against OpenAI, though gave them a chance to amend and refile. Itโ€™s just the latest skirmish in the ongoing divorce proceedings between the AI industry's most famous frenemies.

Source: The Verge

Deep Dives

โšก Cloudflare rebuilds Next.js on Vite in one week

In a flex that's equal parts impressive and terrifying, a Cloudflare engineer used AI to port Next.js to Vite, creating 'vinext'. The result? Builds are 4x faster and bundles are 57% smaller. Itโ€™s a stark reminder that 'legacy' codebases are on notice.

Source: Cloudflare Blog

๐Ÿ•ธ๏ธ Why multi-agent workflows keep faceplanting

GitHub's engineering team dropped a guide on why your agent swarms are failing. Hint: Itโ€™s not the model's fault; itโ€™s your lack of structure. They outline three patterns to stop your AI agents from getting stuck in infinite loops of politeness.

Source: GitHub Blog

๐Ÿ‡ฉ๐Ÿ‡ฐ Denmark says 'Farvel' to Microsoft

The Danish government is making a hard pivot toward digital independence, aiming to ditch Microsoft software to reduce reliance on foreign tech giants. Itโ€™s a bold move that will likely involve a lot of Linux ISOs and very few Clippy assistants.

Source: The Record

Engineering & Research

๐Ÿ˜ The 100 Million Row Challenge... in PHP?

Because apparently we aren't suffering enough, developers are now trying to process 100 million rows of data using PHP. Itโ€™s a fascinating benchmark of modern PHP performance and a great way to heat your home this winter.

Source: GitHub

๐ŸŽฎ LLM Skirmish: RTS for AI Agents

A new environment designed to test if LLMs can actually plan and execute strategies in real-time, rather than just writing poetry about it. If your model can't beat a rush attack, is it really AGI?

Source: LLM Skirmish

๐Ÿ“‰ HyperNova 60B: Compressed but mighty

Spanish startup Multiverse Computing released a compressed 60B parameter model that claims to punch above its weight class. It's another data point in the 'smaller, denser models' trend that's finally making local inference viable.

Source: TechCrunch

Odds & Ends

๐Ÿบ Boozy chimps confirm scientific theory

Researchers analyzed urine from chimps and confirmed they metabolize ethanol, supporting the 'drunken monkey' hypothesis. Yes, our ancestors likely got a buzz from fermented fruit, explaining a lot about human history.

Source: Ars Technica

โ˜€๏ธ Solar finally eclipses Hydro in the US

For the first time, solar power has generated more electricity than hydroelectric dams in the US. The sun is officially pulling its weight, even if the coal plants are still lurking in the background.

Source: Ars Technica

โ† Back to archive