Code Monkey

Coding with Lewis

databricks

Debezium

Docker

ElixirStatus

<h2>Must know</h2>
<ul>
<li><strong>Meta makes massive AI chip deal with AMD</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 shares, contingent on AMD’s stock reaching $600, underscoring the scale of Meta’s commitment to ‘personal superintelligence.’ <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 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, particularly for DRAM, NAND flash, and server CPUs, driven by the AI industry’s demand straining global supply chains. This situation highlights how smaller providers, lacking the long-term vendor contracts of hyperscalers like AWS and Google, are forced to pass these rising costs directly to their users. <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 without relying on cloud processing. This 3B-parameter multimodal model aims for low latency, privacy, and offline capability across mobile, web, and desktop platforms. It achieved 91.6% on the ScreenSpot-V2 GUI grounding benchmark, performing competitively with much larger models, though it struggles with long-horizon multi-step tasks. <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>Quesma releases OTelBench for OpenTelemetry and AI performance</strong>: Quesma released OTelBench, an open-source benchmarking suite to evaluate OpenTelemetry pipeline performance and assess AI agent effectiveness in Site Reliability Engineering (SRE) tasks, revealing LLMs often achieve success rates below 30% in production-grade instrumentation scenarios. <a href="https://app.daily.dev/posts/OB1MlP8PG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google enhances Opal with agentic AI workflows</strong>: Google upgraded its Opal Workflow builder with new agentic capabilities, including an “Agent step” that analyzes goals, determines approaches, and calls tools automatically, powered by Gemini 3 Flash, allowing non-technical users to create complex mini-apps with text prompts. <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>: A new npm supply chain attack, SANDWORM_MODE, targets developer tools and AI coding assistants like Claude Code, using typosquatted packages to harvest npm/GitHub tokens and cloud credentials, and deploying malicious MCP servers for prompt-injection attacks. <a href="https://app.daily.dev/posts/q6c2RLQKH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>DMARCbis update discussion released</strong>: DMARCbis, an upcoming update to the DMARC email authentication standard, introduces clearer reporting expectations, standardized DNS tree walks for receiver-side policy discovery, and record tag updates to reduce ambiguity, as discussed in a fireside chat by Sendmarc. <a href="https://app.daily.dev/posts/yEFoBF8zN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitLab GPG key extended</strong>: GitLab extended its GPG key for repository metadata from February 27, 2026, to February 6, 2028, to ensure integrity and minimize disruption for users, though those who configured repositories before February 17, 2026, must manually update the key. <a href="https://app.daily.dev/posts/JTFdrzyr9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI model performance upgraded in Cursor</strong>: OpenAI models integrated into Cursor IDE were upgraded to use WebSockets with the Responses API, reportedly resulting in up to 30% faster performance for all Cursor users, according to a promotional announcement. <a href="https://app.daily.dev/posts/CNQCNZ2N0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent browser sandbox project announced</strong>: A new project named “browser sandbox” was announced, aiming to create a secure browser environment for AI agents to interact with the web, featuring “zero config” setup, though details are currently truncated and the announcement date is in the future. <a href="https://app.daily.dev/posts/AY0JUdVv6" 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 Friday deadline to Anthropic to remove AI guardrails on Claude for unrestricted military use, threatening to designate them a ‘supply chain risk’ or invoke the Defense Production Act if they refuse. Apparently, the DOD is so dependent on Anthropic for ‘classified frontier AI access’ that they’re willing to throw around threats usually reserved for foreign adversaries. Because nothing says ‘innovation’ like strong-arming your vendors into compromising their ethical stance. <a href="https://app.daily.dev/posts/vbdfUwGoN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft and Salesforce AI conversation analysis announced</strong>: Microsoft Research and Salesforce AI reportedly analyzed over 200,000 AI conversations, confirming ‘industry suspicions’ without providing any specific details about what those suspicions were or what was actually found. This vague social media post is a masterclass in generating curiosity without delivering any actual information. It’s almost as if they want us to believe something profound happened, but can’t quite articulate what that ‘something’ is. <a href="https://app.daily.dev/posts/FAzFsVs8B" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
