📱Big Tech & Startups
Google co-founder Larry Page reportedly has a new AI startup (1 minute read)
Dynatomics is a new company by Google co-founder Larry Page that is focused on applying AI to product manufacturing. The company aims to create highly optimized designs for objects and then have a factory build them. The effort is being run by Chris Anderson, the previous CTO of electric airplane startup Kittyhawk. Other companies exploring the ways AI could be used to improve manufacturing processes include Orbital Materials, PhysicsX, and Instrumental.
SpaceX Starship spirals out of control in second straight test flight failure (3 minute read)
SpaceX's Starship spiraled out of control while in space during a test on Thursday - this is the second launch in a row that the vehicle has run into a fatal problem on its way to orbit. Flights into major Florida airports were briefly halted. The booster was successfully caught for the third time by the launch tower. SpaceX had made improvements to the lines that send fuel to Starship's engines, changed the temperature of the propellent, and added extra vents and a purge system ahead of the recent test flight.
🚀Science & Futuristic Technology
Scientists Discover Thousands of New Microbial Species Thriving in the Mariana Trench (7 minute read)
The Mariana Trench Environment and Ecology Research Project has added over 6,000 new microbes to the number of species found at the deepest regions of the ocean. Its investigations have revealed two adaptations that allow the microbes to thrive under intense pressure and a mutated gene that could boost their ability to survive. The study of these organisms could yield new medications and information that could be useful for engineering pressure- or radiation-resistant proteins.
Intuitive Machines Is Set For Lunar Landing – With Hopping Drone, Ice Drill, And 4G (4 minute read)
Intuitive Machines' Athena moon lander didn't crash, but it may not be upright. The lander was set to descend at Mons Mouton, a plateau near the lunar South Pole, at no earlier than 12:32 PM ET on Thursday. It contains a drill to search for ice, a 4G network test, three rovers, and a first-of-its-kind hopping drone. It is unclear how the imperfect landing may impact the mission. Athena launched last Wednesday aboard a SpaceX Falcon 9 rocket.
💻Programming, Design & Data Science
Mistral adds a new API that turns any PDF document into an AI-ready Markdown file (4 minute read)
Mistral OCR is an optical character recognition API that can turn any PDF into a text file. It is multimodal and can detect when there are illustrations or photos intertwined with blocks of text. The output is formatted into Markdown. Mistral OCR is available on Mistral's own platform and through its cloud partners - Mistral offers on-premise deployment for companies working with classified or sensitive data.
ChatGPT on macOS can now directly edit code (2 minute read)
The newest version of OpenAI's ChatGPT app for macOS can edit code in supported developer tools, including Xcode, VS Code, and JetBrains. Users can turn on an 'auto-apply' mode to allow ChatGPT to make edits without the need for additional clicks. The feature is available to ChatGPT Plus, Pro, and Team subscribers. It will roll out to Enterprise, Edu, and free users next week. The ChatGPT app for Windows will get direct editing soon.
🎁Miscellaneous
A quarter of startups in YC's current cohort have codebases that are almost entirely AI-generated (3 minute read)
While each of the founders that Y Combinator funded in its current cohort is highly technical and completely capable of building their own products from scratch, a quarter of the startups have 95% of their code bases generated by AI. Product builders who rely heavily on AI will need to be good at reading code and finding bugs. VCs and developers have been excited about AI-powered coding, with multiple startups raising hundreds of millions of dollars in funding in the last 12 months.
The Hottest AI Companies Right Now Are ‘Apps' (7 minute read)
AI wrappers are all the rage today - startups are offering AI chatbots, research tools, and other software applications for coding, clinicians, and customer service, all built at least in part on large language models. Investors are eager to put money into these services. However, some have started to question the wisdom of AI companies powering billions into developing models as companies like DeepSeek have proven competitive options can be created for much less. One of the biggest concerns with AI app companies is that a large language model developer may decide to compete directly with them, but it is unlikely that AI developers will build verticalized apps for every single possible use case.
⚡Quick Links
Ask HN: What less-popular systems programming language are you using? (Hacker News Thread)
Some of the languages developers are using other than C, C++, Rust, and Go include Zig, Odin, OCaml, and Nim.
Diffusion models are interesting (2 minute read)
Diffusion large language models start all at once and then gradually come up with sensible words simultaneously - the technique is creating models that outperform similar-sized large language models in code generation.
Alibaba shares soar after Chinese tech giant unveils new DeepSeek rival (3 minute read)
Alibaba's QwQ-32B operates with 32 billion parameters compared to DeepSeek R1's 671 billion parameters - while rivaling its performance.
Sam Altman's World Network Unveils New Chat Feature to Connect Real Humans (3 minute read)
World Chat will help people to discern bots from humans by giving World Network digital passport holders - people who have proof-of-personhood-verified accounts - blue chat bubbles and a unique World ID gem displayed in chats.
Meta is targeting ‘hundreds of millions' of businesses in agentic AI deployment (4 minute read)
Meta's business AIs will be able to automate redundant tasks, help businesses communicate and find more customers, and provide an almost concierge-like service to customers.
Warewulf (GitHub Repo)
Warewulf is a stateless, network-boot cluster provisioning system designed primarily for large clusters of bare metal and virtual systems.