<h2>Must know</h2>
<ul>
<li><strong>Meta goes nuclear for AI infrastructure</strong>: Meta is partnering with Vistra, TerraPower, and Oklo to secure up to 6.6 GW of nuclear energy by 2035. This massive investment aims to power Meta’s expanding AI infrastructure and data centers, particularly the Prometheus supercluster in Ohio. The move highlights the extreme energy demands of advanced AI and Meta’s commitment to reliable, baseload power, as announced by Meta itself. <a href="https://app.daily.dev/posts/elp6Y0W2r" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linus Torvalds uses AI for ‘vibe coding’</strong>: Even Linus Torvalds, the Linux kernel creator, is now using AI coding tools for ‘vibe coding,’ specifically Google’s Antigravity AI-assist IDE (likely Gemini) for a Python visualization tool in his new AudioNoise project. While he still codes the core DSP in C himself, he sees AI as streamlining the process between searching for code examples and manual coding, marking a significant adoption by a major open-source figure. <a href="https://app.daily.dev/posts/fe8LDMARY" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>India’s smartphone security demands spark industry backlash</strong>: India is proposing 83 new smartphone security standards, including source code access and 12-month device log storage, drawing strong opposition from tech giants like Apple, Samsung, and Google. Industry players argue these requirements are unprecedented, invasive, and could compromise proprietary technology. The Indian IT ministry publicly denies seeking source code, but leaked documents suggest otherwise, creating significant regulatory uncertainty, as reported by Reuters. <a href="https://app.daily.dev/posts/C0BEamxdR" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Universal Commerce Protocol aims to unify e-commerce</strong>: Google, Shopify, Walmart, Target, and Etsy have launched the Universal Commerce Protocol (UCP), an open-source standard to streamline product discovery and order management, integrating with AI search and Gemini apps for direct checkout. <a href="https://app.daily.dev/feed-by-ids?id=7CfzB00S6&amp;id=HH05G9maY&amp;id=MhDEwyYbE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>D-Wave acquires Quantum Circuits for gate-model expansion</strong>: D-Wave Quantum is acquiring Quantum Circuits Inc. for $550 million to combine D-Wave’s annealing quantum computing with Quantum Circuits’ error-corrected gate-model technology, aiming for a commercial gate-model system by 2026. <a href="https://app.daily.dev/posts/rdMlkyvEb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>US reauthorizes National Quantum Initiative</strong>: The bipartisan National Quantum Initiative Reauthorization Act has been introduced to extend federal coordination of quantum research through 2034, establishing new NIST quantum centers and expanding NASA quantum programs. <a href="https://app.daily.dev/posts/fWH0q50O1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Motional reboots robotaxi with AI-first approach</strong>: Motional is relaunching its robotaxi service, shifting from rules-based systems to a foundation model-based AI architecture, with employee testing starting in Las Vegas and public ride-hailing targeted for mid-2026. <a href="https://app.daily.dev/posts/JFLFFu1SK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>PostgreSQL as a Redis alternative for specific use cases</strong>: A technical article suggests PostgreSQL can replace Redis for caching, pub/sub, and job queues in small-to-medium apps by leveraging features like UNLOGGED tables and LISTEN/NOTIFY, potentially simplifying infrastructure and saving costs. <a href="https://app.daily.dev/posts/X7d3O8Vu2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Gentoo Linux 2025 review highlights technical milestones</strong>: Gentoo Linux’s 2025 year-in-review reports 112,927 commits, four new developers, finalized EAPI 9, migration planning to Codeberg, and official support for RISC-V bootable images and Windows Subsystem for Linux. <a href="https://app.daily.dev/posts/19Mn2n5YA" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>European Commission consults on open source strategy</strong>: The European Commission has launched a consultation on open source software strategy, aiming to reduce reliance on US ‘Big Tech’ by classifying open source as critical infrastructure and fostering EU developer communities. <a href="https://app.daily.dev/posts/nSTd14JQu" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI code generation: 70-90% of production code, zero data</strong>: A report claims AI code generation tools now create 70-90% of production code, urging developers to adapt or ‘fall behind.’ This bold assertion, lacking any detailed data source or named examples, sounds suspiciously like a FOMO-inducing marketing ploy. Apparently, we should just trust that AI is doing all the heavy lifting now, even if nobody can actually prove it. <a href="https://app.daily.dev/posts/LeN1TdPNx" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents: just bash scripts with extra steps</strong>: Vercel and others are touting ‘minimalist AI agent architectures’ using BASH shells and Unix tools as a breakthrough, claiming they outperform ‘complex, over-engineered systems.’ So, after all the hype about sophisticated AI frameworks, the big secret is… just use a shell script? It seems the ‘agentic success’ they’re talking about is just rediscovering the Unix philosophy, but with more tokens and a fancier name. <a href="https://app.daily.dev/posts/imAYzPlMb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GenAI projects fail because of… lack of basic engineering?</strong>: An anonymous author claims 80% of GenAI projects fail due to ‘lack of production-ready infrastructure,’ not model capability. Apparently, these projects are missing basic engineering practices like config management, error handlers, and prompt versioning. So, the revolutionary AI is failing because people forgot how to build software? Who would’ve thought that throwing a fancy model at a problem doesn’t magically fix bad engineering. <a href="https://app.daily.dev/feed-by-ids?id=ExsMXw27a&amp;id=yocwu10Ql" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
