<h2>Must know</h2>
<ul>
<li><strong>Nvidia and AMD unveil next-gen AI hardware</strong>: Nvidia launched its Vera Rubin AI computing architecture at CES 2026, promising 5x inference performance of Blackwell and 10x lower inference costs. This platform integrates the Vera CPU, Rubin GPU, and next-gen networking chips. Meanwhile, AMD announced its Venice server CPUs with up to 256 cores and MI400 datacenter accelerators featuring 12 HBM4 memory dies, both slated for late 2026, as reported by Phoronix and other CES coverage. These developments signal a significant leap in hardware capabilities for AI workloads. <a href="https://app.daily.dev/feed-by-ids?id=UyBvDFk8k&amp;id=QMRXI1rpB&amp;id=fj4JypEc8&amp;id=pPE45Xr1Z" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft acquires Osmos for AI data engineering in Fabric</strong>: Microsoft has acquired 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, according to various tech news outlets. <a href="https://app.daily.dev/posts/Tm2Js37ee" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents demand stateful infrastructure, challenging serverless</strong>: AI agents are forcing a re-evaluation of cloud architecture, as their need for persistent execution environments to maintain context across multi-step tasks clashes with stateless serverless models like AWS Lambda. Enterprises are increasingly shifting AI workloads to VPCs and on-premises infrastructure for enhanced security and compliance, marking a departure from a decade-long trend toward ephemeral cloud solutions. This shift requires rethinking orchestration, resource allocation, and governance for stateful applications, as highlighted by The New Stack. <a href="https://app.daily.dev/feed-by-ids?id=7y67CNGeP&amp;id=D5zAPFRRJ&amp;id=SXn4AkZZF" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Elixir ecosystem thrives with security and AI advancements</strong>: The Elixir ecosystem is seeing continued growth, with developers migrating from JavaScript frameworks like Next.js to Phoenix/Ash due to improved supply chain security and stability, alongside advancements in AI/ML integrations and enhanced type systems. <a href="https://app.daily.dev/feed-by-ids?id=AIHWiBo8m&amp;id=eLUei9iM4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft and Canonical boost AI infrastructure support</strong>: Microsoft is integrating Osmos’ AI data engineering into Fabric, while Canonical announced Ubuntu support for NVIDIA’s Rubin platform and Vera Rubin NVL72 rack-scale systems, enhancing Arm architecture support and simplifying LLM deployment. <a href="https://app.daily.dev/feed-by-ids?id=Tm2Js37ee&amp;id=2aoVYu5cf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered web scraping tools emerge at OxyCon 2025</strong>: New AI tools like Cursor and Oxylabs’ AI Studio are enabling large-scale web scraping using natural language, generating self-healing parsers that adapt to web page changes, though human expertise remains crucial for anti-bot strategies. <a href="https://app.daily.dev/feed-by-ids?id=3Nqb1wXgV&amp;id=3TA62DRee&amp;id=Wqsx0Gsbo" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Slack overhauls infrastructure, bug bounty platform faces scrutiny</strong>: Slack has restructured its Chef-based configuration management to enhance deployment safety and reduce failure impact, while HackerOne faces criticism for delayed payments to security researchers, highlighting operational strains in bug bounty programs. <a href="https://app.daily.dev/feed-by-ids?id=bBQsTOLyp&amp;id=bbTfrQofX" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New AI models optimize performance and reasoning</strong>: ByteShape demonstrated its Shapelearn method running a 30B parameter Qwen3 model on a Raspberry Pi 5 at over 8 tokens per second, while DeepSeek released DeepSeek-V3.2, an open-source reasoning model claiming superior performance against GPT-5 on benchmarks. <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>AWS ECS gains tmpfs support for ephemeral data</strong>: Amazon ECS now supports tmpfs mounts for Linux tasks on both AWS Fargate and ECS Managed Instances, providing in-memory file systems for fast, ephemeral data like secrets or caches, enhancing security and performance. <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>GNOME and Firefox consider disabling middle-click paste</strong>: GNOME and Firefox are debating disabling the X11 legacy middle-click paste feature by default, citing poor discoverability and accidental activation, despite strong opposition from power users who rely on it for productivity. <a href="https://app.daily.dev/posts/bH8ngfo5B" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI-generated code: more commits, more problems?</strong>: One developer claims 100% of their code contributions for the past month were AI-generated, boasting 259 PRs and 40,000+ lines. Yet, the 2026 Cortex benchmark conveniently shows a 23.5% increase in incident PRs for teams using AI tools, with higher failure rates and longer resolution times. So, while AI might make you feel like a coding god, it seems it’s also making everyone else clean up the mess. <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>AI companion robots: because your desk needs a personality crisis</strong>: CES 2026 was apparently all about AI companion robots that prioritize ‘emotional connection’ over actual utility. Loona’s DeskMate, for instance, offers ‘expressive displays’ and ‘animated personalities.’ Because what every busy tech professional needs is a desk ornament having an existential crisis, rather than, you know, a robot that actually cleans the house or something useful. Specialized robots are still doing the real work, but hey, at least your desk can now judge your emotional state. <a href="https://app.daily.dev/feed-by-ids?id=1orFwq3Up&amp;id=Ixfjkj22m&amp;id=T4ITTdiGL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Lenovo’s AI glasses: concept or just a fancy paperweight?</strong>: Lenovo showcased ‘concept’ AI glasses at CES 2026, boasting a 2MP camera, 214mAh battery, and features like live translation and image recognition. The catch? They were a non-working prototype with no demonstrated functionality, release date, or pricing. So, essentially, Lenovo unveiled a pair of lightweight frames with some impressive promises, proving that in the tech world, a good concept is often just a glorified sketch with a hefty marketing budget. <a href="https://app.daily.dev/feed-by-ids?id=iobV97iR6&amp;id=oQMoRDfAV" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
