<h2>Must know</h2>
<ul>
<li><strong>React moves to an independent foundation</strong>: Meta has officially transferred React, React Native, and the JSX specification to the newly formed React Foundation under the Linux Foundation. While Meta remains a platinum member alongside Microsoft and Vercel, technical direction stays with the existing maintainers to ensure community-led growth. This shift aims to professionalize funding and global accessibility for the ecosystem. <a href="https://app.daily.dev/posts/Vf1iZDHHt" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic acqui-hires Vercept to fix Claude’s computer use</strong>: Anthropic has acquired the Seattle startup Vercept, the team behind the ‘Vy’ agent that could remotely operate MacBooks. The acquisition is a clear move to bolster Claude’s ability to interact with desktop GUIs, though Vercept’s standalone product will be shuttered in March. The deal was reportedly marked by a public spat between investors over the early exit strategy. <a href="https://app.daily.dev/posts/VoPayj2dv" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Nvidia hits record 215 billion annual revenue</strong>: Nvidia reported a staggering 73% year-over-year revenue increase, driven almost entirely by its $62 billion quarterly data center business. CEO Jensen Huang dismissed concerns about AI capex sustainability, noting that even six-year-old GPUs are currently fetching high prices due to compute token demand. The company also confirmed ongoing $30 billion partnership talks with OpenAI. <a href="https://app.daily.dev/posts/I58Ho2pR9" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Google launches Gemini-powered Android automation</strong>: New AppFunctions APIs allow Gemini to autonomously operate apps like ride-sharing and food delivery on Pixel 10 and Galaxy S26 devices. <a href="https://app.daily.dev/posts/hXcElb2hp" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Mercury 2 LLM targets real-time latency</strong>: Inception’s new model uses parallel refinement and diffusion-based reasoning to generate multiple tokens simultaneously for lag-free agentic loops. <a href="https://app.daily.dev/posts/ea5kyYvcg" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cilium 1.19 hardens Kubernetes networking</strong>: The 10th-anniversary release introduces strict IPsec/WireGuard modes and sidecar-free mutual authentication via Ztunnel integration. <a href="https://app.daily.dev/posts/VJq8q5iXQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Laravel AI SDK benchmarked across five providers</strong>: DeepSeek and Grok emerged as the most cost-effective for text, while GPT-4.5 surprisingly trailed GPT-4o mini in certain speed tests. <a href="https://app.daily.dev/posts/TjfKXRrSQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Unity kills HDRP in favor of unified URP</strong>: Unity is moving HDRP to maintenance mode to focus all resources on URP, which will now receive features like real-time global illumination. <a href="https://app.daily.dev/posts/kOn1P0sth" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>npm 11.10 introduces release age gating</strong>: A new minimumReleaseAge setting allows teams to block brand-new packages, providing a buffer against immediate supply chain attacks. <a href="https://app.daily.dev/posts/Tw6GntfNF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Warper 7.2 brings Rust-powered virtualization to React</strong>: This library offloads scroll calculations to WebAssembly, claiming 120 FPS for lists with 100,000 items. <a href="https://app.daily.dev/posts/SvL1JwjLf" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Google Research claims a magical 76 percent accuracy boost</strong>: A single unverified retweet claims Google boosted LLM accuracy by 76 points with “zero extra tokens” and “zero latency.” Without a paper or methodology, this sounds like a breakthrough in the field of wishful thinking. We’ll wait for the actual math before we start deleting our prompt engineering guides. <a href="https://app.daily.dev/posts/73tgOim8I" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Salesforce invents Agent Work Units to hide modest growth</strong>: Marc Benioff is trying to pivot the ‘SaaSpocalypse’ narrative by introducing ‘Agent Work Units’ as a metric for AI productivity. It’s a convenient distraction from the fact that their 13% revenue growth was largely propped up by an $8 billion acquisition. Apparently, if you can’t sell more software, you just invent a new way to count the work the software isn’t doing yet. <a href="https://app.daily.dev/posts/R4O8qdc3C" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The rise of Potemkin codebases</strong>: AI agents are flooding open source with ‘slop’ pull requests, leading to codebases that look active but lack any human with a mental model of the system. Maintainers like Daniel Stenberg are dropping bug bounties because AI-generated reports have diluted useful signal from 15% down to 5%. We’re effectively trading long-term system integrity for a high-velocity stream of locally coherent garbage. <a href="https://app.daily.dev/feed-by-ids?id=G15hNCZJZ&amp;id=NYF36mKyy" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
