<h2>Must know</h2>
<ul>
<li><strong>Nvidia and AMD unveil next-gen AI hardware at CES 2026</strong>: Nvidia launched its Vera Rubin AI computing architecture, featuring the Vera CPU and Rubin GPU, promising 5x inference performance over Blackwell and 10x lower inference costs. They also introduced Spectrum-X Ethernet Photonics for AI data centers, claiming 5x power reduction per port. AMD countered with Venice server CPUs, packing up to 256 cores, and MI400 datacenter accelerators with HBM4 memory, both slated for late 2026. These announcements, heavily covered by sources like Phoronix and The New Stack, signal an intense arms race in AI infrastructure, with both companies pushing the boundaries of chip design and networking for massive AI workloads. <a href="https://app.daily.dev/feed-by-ids?id=UyBvDFk8k&amp;id=DyqaVJ0CP&amp;id=ZF9igCOMr&amp;id=GnVIXciCG&amp;id=QMRXI1rpB&amp;id=fj4JypEc8&amp;id=pPE45Xr1Z" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI model optimization and new reasoning benchmarks emerge</strong>: ByteShape demonstrated its Shapelearn method, enabling a 30B parameter Qwen3 model to run on a Raspberry Pi 5 at over 8 tokens per second, optimizing weight datatypes for throughput and accuracy. DeepSeek released DeepSeek-V3.2, an open-source reasoning and agentic AI model family, claiming superior performance over GPT-5 and parity with Gemini-3.0-Pro on several benchmarks, featuring DeepSeek Sparse Attention for longer context windows. These developments, reported by sources like ByteByteGo, highlight a dual focus in AI: making large models efficient on consumer hardware and enhancing reasoning capabilities for complex tasks. <a href="https://app.daily.dev/feed-by-ids?id=2chvqxrcS&amp;id=YXZIO7mia" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft acquires Osmos for AI data engineering in Fabric</strong>: Microsoft announced the acquisition of Osmos, an AI-powered data engineering platform, to integrate its technology into Microsoft Fabric. Osmos specializes in agentic AI for automating ETL tasks, transforming raw data into analytics-ready assets, and its AI agents will support Fabric’s OneLake. This move aims to unify and simplify data and analytics workflows, focusing on AI-readiness, and could intensify competition with existing Azure partners like Databricks, as noted by multiple tech outlets. <a href="https://app.daily.dev/posts/Tm2Js37ee" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Elixir ecosystem gains traction with focus on stability and security</strong>: The Elixir ecosystem saw significant growth in 2025, with developers migrating from JavaScript frameworks like Next.js to Phoenix/Ash due to supply chain concerns and improved safety, achieving feature parity deployments in weeks. <a href="https://app.daily.dev/posts/AIHWiBo8m" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks launches AI-powered debugging for OLTP databases</strong>: Databricks introduced an AI-powered agentic debugging platform that reduces OLTP database debugging time by up to 90% across thousands of instances, evolving from anomaly detection to an interactive chat assistant with multi-agent collaboration. <a href="https://app.daily.dev/posts/xdHmgnzKW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated code raises productivity and incident rates</strong>: A Claude Code developer claimed all their code contributions for 30 days were AI-generated, while the 2026 Cortex benchmark reported a 23.5% increase in incident PRs for teams using AI tools, with higher failure rates and longer resolution times. <a href="https://app.daily.dev/feed-by-ids?id=XLHgNaGC9&amp;id=w3uV6bVaD" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Serverless architectures struggle with stateful AI agents</strong>: Serverless architectures like AWS Lambda are proving inadequate for production AI agent workflows due to their stateless nature, forcing enterprises to shift towards VPC and on-premises deployments for persistent execution environments and context retention. <a href="https://app.daily.dev/posts/SXn4AkZZF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Gentoo Linux makes progress on RISC-V and WSL with minimal funding</strong>: Gentoo Linux’s 2025 retrospective highlighted achievements like RISC-V bootable VM images and weekly WSL images, all while operating on a modest annual income of $12,066 and migrating from GitHub to Codeberg to avoid Microsoft Copilot. <a href="https://app.daily.dev/posts/Kad9BYy6h" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New AWS ECS tmpfs support enhances container performance and security</strong>: Amazon ECS now supports tmpfs mounts for Linux tasks on both AWS Fargate and ECS Managed Instances, enabling in-memory file systems for fast, ephemeral data and supporting read-only container root filesystems for improved security. <a href="https://app.daily.dev/feed-by-ids?id=XLHgNaGC9&amp;id=w3uV6bVaD" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Python library simplifies AWS CloudFormation template creation</strong>: Troposphere, a Python library, allows developers to create AWS CloudFormation templates using Python code instead of JSON, offering type checking, property validation, and early error detection for both AWS and OpenStack resources. <a href="https://app.daily.dev/posts/9UOzojnGl" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI will augment, not replace, workers (unless you’re entry-level)</strong>: OpenAI’s research, as cited in a podcast, confidently predicts AI will augment most human workers, not replace them. However, an ex-Uber recruiter casually mentions AI is already reducing some engineering roles, particularly affecting developer velocity teams, and the bar for entry-level positions has risen significantly. So, while the AI overlords promise augmentation, it seems the ‘augmentation’ for new grads might just be the augmented difficulty of finding a job. <a href="https://app.daily.dev/feed-by-ids?id=v2rIwoMpZ&amp;id=zv61ZRYsN&amp;id=lQ3L56bdN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered web scraping: the future is now (and still needs humans)</strong>: OxyCon 2025 proudly showcased AI tools that let you generate web scrapers with natural language, treating web pages like structured APIs. They claim it’s ‘fundamentally changed workflow’ and ‘eliminates the need to hire engineers for simple scraping tasks.’ Yet, in the same breath, they admit human expertise is ‘crucial’ for anti-bot defenses and that AI ‘cannot yet fully replace expert decision-making.’ So, it’s revolutionary, but also, you still need to pay the experts. Got it. <a href="https://app.daily.dev/feed-by-ids?id=3Nqb1wXgV&amp;id=3TA62DRee&amp;id=Wqsx0Gsbo&amp;id=AyHGRwwVX" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Lenovo’s AI-powered laptop: because your screen needs to follow your head</strong>: Lenovo unveiled the ThinkBook Plus Gen 7 Auto Twist, a laptop with a motorized pivoting hinge that uses head tracking to adjust the screen angle. Because apparently, manually adjusting your screen is too much effort, and your laptop needs to follow your every move. It also has an AI companion and live translation, so your laptop can awkwardly pivot while you try to have a conversation. Innovation, or just a solution looking for a problem? <a href="https://app.daily.dev/feed-by-ids?id=iobV97iR6&amp;id=oQMoRDfAV" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
