Must know

RAM and storage prices skyrocket due to AI demand: Memory and storage prices have surged 2-3x in early 2026, primarily driven by insatiable AI data center demand for high-bandwidth memory (HBM). The transition away from DDR4 and Micron’s exit from the consumer market are exacerbating supply constraints, with NAND flash prices doubling. Kingston reports a 246% increase in wafer prices, and forecasts suggest another 50% hike for DDR5 and SSDs in the first half of 2026, impacting both consumers and PC manufacturers. Read more
Meta secures nuclear energy for AI data centers: Meta has partnered with Vistra, TerraPower, and Oklo to secure up to 6.6 GW of nuclear power by 2035 for its AI infrastructure and data centers. This initiative involves extending the life of existing plants and developing advanced reactors, with a primary focus on the Prometheus supercluster in Ohio. The move aims to provide clean, reliable baseload power, supporting Meta’s AI expansion and generating thousands of jobs, as announced by Meta. Read more
Amazon’s AI ‘buy for me’ feature sparks merchant outrage: Amazon’s new AI-driven ‘Buy For Me’ feature is scraping independent merchant websites and listing their products on Amazon without consent, leading to significant backlash. Small business owners report surprise orders, incorrect product images, and public exposure of wholesale pricing. Merchants are criticizing Amazon’s opt-out mechanism and the ethical implications of using AI scraping while blocking others, highlighting a power imbalance in AI-powered commerce. Read more

Good to know

Linus Torvalds uses AI for new open-source project: Linus Torvalds released AudioNoise, an open-source digital guitar pedal emulator, using Google Antigravity for AI-assisted ‘vibe coding’ on the Python visualization component, showcasing practical AI integration in hobbyist development. Read more
Python 3.14 promises speed and threading improvements: Python 3.14, slated for 2026, will introduce lazy imports and official free-threading, potentially offering up to 10x speedups for parallel workloads, alongside a growing trend of Rust-based type checkers and packaging tools like uv and Hatch. Read more
Dell GB10 edge AI platform leverages Nvidia Blackwell: Dell’s GB10 device, featuring Nvidia’s Blackwell GB10 superchip, is optimized for edge AI, running large language models up to 120 billion parameters with 128GB unified memory, enabling low-latency, privacy-focused AI without cloud reliance. Read more
EU open source strategy targets big tech dependence: The European Commission is consulting on an open source strategy to reduce reliance on U.S. tech giants, treating open source as critical infrastructure and shifting funding to scale viable solutions in cloud, AI, and cybersecurity. Read more
AWS launches VPC encryption controls: AWS introduced VPC Encryption Controls to monitor and enforce in-transit encryption within and between VPCs, offering monitor and enforce modes, with the latter requiring Nitro-based infrastructure and supporting compliance efforts for HIPAA, PCI DSS, and FedRAMP. Read more
Google and partners launch universal commerce protocol for AI agents: Google, with Shopify, Walmart, Target, and Etsy, launched the Universal Commerce Protocol (UCP), an open-source standard enabling AI agents to manage the full purchasing funnel from discovery to checkout, integrating with existing payment systems and Google’s AI search. Read more
Docker releases free open-source hardened images: Docker has released over 1,000 hardened, open-source images under the Apache 2.0 license, designed to improve container security by removing unnecessary packages and outdated dependencies, alongside hardened Helm charts for Kubernetes deployments. Read more
Percona Everest launches as open source cloud-native database platform: Percona Everest is a new open-source, cloud-native platform for deploying and managing PostgreSQL, MongoDB, and MySQL clusters on Kubernetes, offering a UI, automated backups, monitoring, and scaling via Helm charts. Read more
Kilo AI coding agent platform launches with broad IDE integration: Kilo, an open-source AI coding agent platform, offers integration with major IDEs, access to over 500 AI models, cloud agents, automated code reviews, and a pay-as-you-go model, aiming to enhance developer productivity and team collaboration. Read more
Gentoo Linux 2025 year in review: Gentoo Linux saw 112,927 commits in 2025, welcomed four new developers, finalized EAPI 9, announced migration to Codeberg, released RISC-V bootable images, and introduced official WSL support, expanding its repository to over 31,000 ebuilds. Read more
Go coding kata repository targets advanced Go practices: The ‘MedUnes/go-kata’ repository offers daily coding challenges for experienced developers to master idiomatic Go patterns, focusing on advanced topics like context-aware concurrency, memory optimization, and robust error handling. Read more
Mac emulation achieves new speeds on ESP32-P4 microcontroller: The Basilisk II emulator has been ported to the ESP32-P4 microcontroller, achieving 15 frames per second for Mac OS 8.1 on a 68040-based Macintosh, a significant leap for microcontroller-based emulation. Read more
Ripple TypeScript UI framework launches: Ripple is a new TypeScript UI framework focusing on simplicity and low boilerplate, featuring unique reactivity with track() and @ syntax, direct DOM updates without a Virtual DOM, and compiler optimizations like dead CSS removal. Read more
LLMOps course expands to attention mechanisms and production challenges: A new segment of the LLMOps course covers attention mechanisms, transformer architectures, mixture-of-experts, and the fundamentals of pretraining and fine-tuning, with hands-on code for production-ready LLM applications. Read more
Google launches free AI-focused upskilling platform: Google Skills, a new free online learning platform, offers over 3,000 courses primarily focused on AI and technology, including prompt engineering, TensorFlow, and generative AI, with personalized paths and Google Career Certificates. Read more
ALPM project completes Arch Linux package management overhaul: The ALPM project, funded by Sovereign Tech Fund, completed a 15-month overhaul of Arch Linux package management, delivering a Rust-based framework, formal specifications, Python bindings, a linting framework, and a new OpenPGP signature verification system. Read more
China accelerates AI sector amid constraints: China’s AI sector is intensifying efforts, with startups MiniMax and Zhipu AI completing successful IPOs, despite challenges like limited access to advanced chip-making equipment and computing infrastructure, driving innovation through co-designing algorithms and hardware. Read more
U.S. quantum technology leadership at risk: An NSF-funded ERVA report warns the U.S. risks losing quantum technology leadership to China and the EU without coordinated investment in quantum materials, biology, computing, and AI, urging a transition from lab demonstrations to manufacturing-ready technologies. Read more
Quantum tech hub advances to EDA awards final stage: The Bloch Quantum, led by Chicago Quantum Exchange, has reached the final stage of the U.S. EDA awards competition, proposing to secure the U.S. quantum technology supply chain and accelerate commercialization across Illinois, Wisconsin, and Indiana. Read more
Encrypted quantum information backup method developed: Researchers at the University of Waterloo have developed a method to duplicate encrypted quantum information without violating the no-cloning theorem, using expiring one-time encryption keys to enable secure, redundant quantum cloud storage. Read more
D-Wave acquires Quantum Circuits Inc.: D-Wave Quantum is acquiring Quantum Circuits Inc. for $550 million, combining D-Wave’s annealing platforms with Quantum Circuits’ error-corrected gate-model technology, aiming to release a commercial gate-model quantum system in 2026. Read more
Linux adds RISC-V safeguard against Spectre V1 attack: The Linux kernel has merged new security patches for RISC-V processors to protect against Spectre V1 microarchitectural attacks, using array_index_nospec() in syscalls to prevent branch predictor poisoning, bringing RISC-V protections in line with x86_64 and ARM64. Read more
Electromagnetic side-channel attack on iris recognition systems demonstrated: Researchers from Shandong University developed EMIRIS, a method to extract biometric iris information from Near-Infrared (NIR) sensor electromagnetic emissions, achieving a 53.47% spoofing success rate on commercial iris recognition systems, exposing significant privacy risks. Read more
Chevereto v4.4 releases S3 and multi-user support for free: Chevereto, a self-hosted photo gallery software, released v4.4, making S3 storage and multi-user support free, positioning itself as a privacy-focused alternative to Imgur and Flickr, with Docker deployment and 2FA now included. Read more
Displace tool translates Docker Compose to Kubernetes: Displace is a new CLI tool that translates Docker Compose configurations into Kubernetes deployments, aiming to simplify migration for developers familiar with Compose but new to Kubernetes, demonstrated by migrating a Laravel app to production Kubernetes in an afternoon. Read more
Ferrite text editor overview and capabilities: Ferrite is a Rust-based text editor using the egui framework, focused on Markdown, JSON, YAML, and TOML, featuring WYSIWYG Markdown editing, syntax highlighting for 40+ languages, MermaidJS diagram rendering, and integrated Git version control. Read more
Hongdown markdown formatter enforces style conventions: Hongdown is a Rust-written Markdown formatter that enforces Hong Minhee’s style conventions, offering CLI functionality for formatting, checking, and diffing, with support for Setext-style headings, 4-space indentation, and 80-column line wrapping. Read more
Legacy rendering tricks in Max Payne (2001): The 2001 video game Max Payne leveraged particle effects, prebaked lightmaps, detail textures, and fake geometry to achieve impressive visuals on limited hardware (450 MHz CPU, 16 MB GPU), demonstrating clever ‘smoke and mirrors’ techniques. Read more
Modernizing 1990s HP vector network analyzer with LCD upgrade: A hardware enthusiast upgraded a 1990s HP 8714C Vector Network Analyzer by replacing its original CRT with a modern color LCD, demonstrating the feasibility of extending the life of robust legacy test equipment. Read more
Intel 8087 FPU microcode conditional tests and design insights: Ken Shirriff’s reverse engineering of the Intel 8087 FPU reveals its microcode engine as a specialized CPU handling complex math and conditional tests directly in microcode, offering enduring lessons for modern ASIC development despite its obsolescence. Read more
Persistence of Visual Basic 6 in business software infrastructure: Visual Basic 6, released in the late 1990s, remains widely used for critical business applications due to its ease of development and deep integration, despite Microsoft’s incompatible VB.NET successor, highlighting technological inertia and migration costs. Read more
Experimental use of Java annotations as executable code via Annote interpreter: Annote is an experimental interpreter allowing Java programs to be written solely with annotations like @Var, @If, @Loop, and @Print, demonstrating a non-standard use of Java’s annotation mechanism for metadata processing, though not intended for production. Read more
Julia GPU ODE solvers significantly outperform JAX and PyTorch: Julia’s GPU-accelerated ODE solvers, particularly the DiffEqGPU library, demonstrate 20x-100x faster performance than JAX and PyTorch alternatives in scientific computing workloads, attributed to Julia’s fundamental architectural approach to GPU utilization and multi-GPU parallelism. Read more
Silent failures in background job systems result in undetected revenue loss: Background job workers can fail silently, leading to revenue loss and fulfillment delays despite monitoring tools showing systems as healthy, necessitating ‘active health checks’ or ‘pulse checks’ for every job worker to detect unprocessed job backlogs. Read more
Engineering management weekly focuses on AI, culture, and organizational practices: A newsletter for engineering managers covers topics like organizational culture debt, the impact of AI on engineering skills, managing cross-team dependencies, strategic focus for teams, and career frameworks for 2026, emphasizing key takeaways for leadership. Read more
Organizational focus yields measurable results: The article argues that organizational focus, achieved by deliberately stopping most activities to concentrate on high-impact priorities, can lead to three times better results in sales, customer satisfaction, and product development, though no evidence is cited. Read more
Relationship between trust and competence in organizational performance: The ‘Slipstream of Trust’ model suggests that optimal competence growth occurs when individuals are given slightly more trust than their current skills merit, asserting that low-trust environments hinder skill development and breed incompetence. Read more
Jakarta EE 12 milestone 2 approaching: Jakarta EE 12 Milestone 2 is set for January 27, 2026, with significant work remaining on specifications, and contributors are encouraged to assist to ensure timely completion, as tracked on the GitHub planning board. Read more
Ask HN: What are you working on? (January 2026): Hacker News users are sharing projects like AI chat tools with session management, a Rust and Tantivy-based full-text search for Claude and Codex sessions, and a Tmux-based CLI tool enabling AI agents to interact with terminal scripts. Read more
PHP to Rust via Copilot: A developer successfully used GitHub Copilot to convert a PHP Laravel Artisan command for scraping TikTok live stream data into Rust code, highlighting AI’s utility in learning new languages and leveraging existing knowledge. Read more
Kotlin Weekly #493: This week’s Kotlin news includes Media3 1.9.0, articles on Kotlin language quirks, Block’s migration to Ktfmt and Detekt, a comprehensive dependency injection guide, Ktor Client pipeline internals, database migrations with Exposed, and new libraries like Carbon Compose and Jindong. Read more
The static dynamic JVM – a many layered dive #JVMLS: John Rose’s JVMLS 2025 talk explores how the JVM balances static analysis with dynamic execution, delving into theoretical computer science and practical optimization techniques that contribute to its power and efficiency. Read more
What’s wrong with multiple h1 elements?: While technically valid, using multiple h1 elements is not recommended for user experience and accessibility, as screen readers rely on a single h1 to describe the primary subject of a page, making multiple h1s confusing. Read more
GYESME: A new design-led downstream of GN: