<h2>Must know</h2>
<ul>
<li><strong>Meta goes all in on AMD for AI chips</strong>: Meta has signed a multiyear deal with AMD, reportedly worth up to $100 billion, to acquire MI540 GPUs and CPUs. This massive investment aims to add 6GW of compute capacity to accelerate AI model development, diversifying Meta’s AI compute stack beyond Nvidia. The agreement includes a performance-based warrant for AMD to issue Meta up to 160 million shares, contingent on AMD’s stock reaching $600. <a href="https://app.daily.dev/posts/vYZNsZv8f" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Hetzner announces significant price hikes due to AI demand</strong>: Hetzner, a major European budget cloud provider, is increasing prices by up to 50% starting April 1, 2026, across all regions. The company attributes these hikes to dramatic increases in hardware costs, specifically DRAM, NAND flash, and server CPUs, driven by the AI industry’s demand straining global supply chains. This highlights how smaller providers, lacking long-term vendor contracts, are forced to pass these costs directly to users, unlike hyperscalers. <a href="https://app.daily.dev/posts/van1jyyDC" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Apple unveils on-device UI AI model</strong>: Apple researchers introduced Ferret-UI Lite, a lightweight, on-device AI model designed to understand and interact with user interfaces directly on mobile, web, and desktop platforms. This 3B-parameter multimodal model prioritizes low latency, privacy, and offline capability, aiming to enable visual perception and control of UIs without relying on cloud processing. It achieved 91.6% on the ScreenSpot-V2 GUI grounding benchmark, competitive with much larger models, and could reduce Siri’s dependence on Google Cloud. <a href="https://app.daily.dev/posts/qj9GoBvQx" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Google enhances Opal with agentic AI workflows</strong>: Google upgraded its Opal Workflow builder with new agentic capabilities, allowing users to create automated workflows using text prompts powered by Gemini 3 Flash. <a href="https://app.daily.dev/posts/ZbbWKhN8A" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Npm supply chain attack targets developer tools and AI</strong>: Socket researchers uncovered an active npm supply chain attack campaign, SANDWORM_MODE, involving 19 typosquatted packages impersonating developer utilities and AI coding tools like Claude Code. <a href="https://app.daily.dev/posts/q6c2RLQKH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Quesma releases OTelBench for OpenTelemetry and AI performance</strong>: Quesma released OTelBench, an open-source benchmarking suite to evaluate OpenTelemetry pipeline performance and AI agent effectiveness in SRE tasks, revealing LLMs often achieve success rates below 30% in production-grade scenarios. <a href="https://app.daily.dev/posts/OB1MlP8PG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Node.js 25.7.0 released with SQLite module RC</strong>: Node.js 25.7.0 has been released, featuring HTTP/2 <code>http1Options</code>, ESM entry points for Single Executable Applications, and a built-in SQLite module marked as a release candidate. <a href="https://app.daily.dev/posts/K4geCuziz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Elasticsearch v9 brings 4x performance boost for vector search</strong>: Trendyol achieved a 4x performance boost by upgrading to Elasticsearch v9.2.1 for vector search, demonstrating significantly higher throughput and lower latency compared to v8.18.3. <a href="https://app.daily.dev/posts/Iw8wLbMFe" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Formae IaC platform expands with multi-cloud support</strong>: Platform Engineering Labs updated formae, its open-source IaC platform, with beta support for GCP, Azure, OCI, and OVHcloud, and a redesigned Plugin SDK for AI-assisted development. <a href="https://app.daily.dev/posts/oAyNaPkCA" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitLab GPG key extended to 2028</strong>: GitLab extended its GPG key for repository metadata from February 27, 2026, to February 6, 2028, to ensure integrity and minimize user disruption. <a href="https://app.daily.dev/posts/JTFdrzyr9" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Pentagon-Anthropic AI standoff over military use</strong>: The Pentagon issued a deadline to Anthropic to remove AI guardrails on Claude for unrestricted military use, threatening to label them a ‘supply chain risk’ or invoke the Defense Production Act. Because apparently, the DOD, despite a directive to avoid single-vendor reliance, is now completely dependent on Anthropic for ‘classified frontier AI access.’ Whoops. <a href="https://app.daily.dev/posts/vbdfUwGoN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI models in Cursor get a ‘30% faster’ upgrade</strong>: OpenAI models integrated into Cursor were upgraded, reportedly resulting in ‘up to 30% faster performance’ for all users by utilizing WebSockets with the Responses API. Because nothing says ‘significant performance boost’ like a vague percentage and a promotional tweet from @leerob, without any actual independent verification. <a href="https://app.daily.dev/posts/CNQCNZ2N0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft and Salesforce confirm ‘industry suspicions’ about AI</strong>: Microsoft Research and Salesforce AI ‘analyzed over 200,000 AI conversations’ and ‘confirmed industry suspicions regarding AI.’ The details? Non-existent. The source? A retweet. This is less a news story and more a clickbait headline designed to generate curiosity without delivering any actual information. Groundbreaking. <a href="https://app.daily.dev/posts/FAzFsVs8B" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
