<h2>Must know</h2>
<ul>
<li>AI coding assistants increase planning demands and security risks: AI coding assistants, while boosting code writing speed, demand significantly more upfront architectural planning and detailed specifications. A 2025 study found developers using AI took 19% longer to complete tasks overall, and code review times increased by 91% for high-AI-usage teams. Furthermore, 45% of AI-generated code contains security vulnerabilities, highlighting the need to treat AI as a junior developer requiring clear direction, not an autonomous agent. This is according to a report from The Miners. <a href="https://app.daily.dev/posts/pYjmmMIkn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>TSMC capital expenditure strategy creates AI chip supply bottleneck: TSMC’s flat capital expenditures from 2021-2024, despite a surge in hyperscaler AI investments, have created a significant supply bottleneck for AI chips. This imbalance is projected to cost hyperscalers billions in lost AI revenue, with shortages expected to persist through at least 2028. The situation suggests a need for hyperscalers to diversify their supply chains, as current TSMC investment plans are insufficient to meet demand. <a href="https://app.daily.dev/posts/pPxEIulIE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Microsoft launches Maia 200 AI inference accelerator: Microsoft has unveiled the Maia 200, a custom AI inference accelerator built on TSMC’s 3nm process. It features native FP8 and FP4 tensor cores, 216GB HBM3e memory, and claims over 10 petaFLOPS in FP4 precision. Microsoft states it offers 30% better performance per dollar than current hardware and will be used for GPT-5.2 models and Microsoft 365 Copilot, with initial deployment in their US Central datacenter. <a href="https://app.daily.dev/posts/HBgJElvHn" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li>Docker Sandboxes introduces microVM isolation for AI coding agents: Docker Sandboxes now offers microVM-based isolation for AI coding agents like Claude Code and Gemini CLI, allowing them to install packages and modify files without affecting the host system, currently available on macOS with Windows and Linux support planned. <a href="https://app.daily.dev/posts/6YdOupFK7" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>OpenAI upgrades ChatGPT code execution with Bash and package installation: OpenAI’s ChatGPT code execution feature now supports running Bash commands and installing packages via pip and npm within its sandboxed environment, expanding its utility for developers across multiple programming languages. <a href="https://app.daily.dev/posts/7V1v7MpYn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Google launches optimized mobile/edge LLM for function calling: Google has released FunctionGemma, a 270M parameter variant of the Gemma 3 model, specifically optimized for efficient function calling on mobile and edge devices, translating natural language into structured API calls for on-device task automation. <a href="https://app.daily.dev/posts/DFyxBQQFb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AMDGPU Linux driver adds HDMI VRR and ALLM gaming support: New AMDGPU Linux kernel driver patches introduce HDMI Variable Rate Refresh (VRR) and Auto Low Latency Mode (ALLM) for gaming on HDMI displays, aiming for feature parity with DisplayPort and Windows platforms despite HDMI Forum restrictions. <a href="https://app.daily.dev/posts/bsF1QAXcQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Uncloud Kubernetes alternative gains momentum among small teams: Uncloud, an open-source deployment platform built on Docker, Caddy, and WireGuard, is gaining traction as a simpler Kubernetes alternative for small teams, offering single-command deployments from source to production without container registries. <a href="https://app.daily.dev/posts/FrS1fa85P" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Linux 6.19 scheduler performance changes and NEXT_BUDDY feature rollback: The Linux 6.19 kernel’s initial test versions experienced performance regressions with the NEXT_BUDDY feature for the EEVDF scheduler, leading to its disabling in the rc7 release candidate after community feedback and benchmarks showed drops in MySQL and other workloads. <a href="https://app.daily.dev/posts/85ivk605B" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Cisco integrates eBPF to modernize firewall and security architecture: Cisco is integrating eBPF technology, acquired through Isovalent, into its enterprise networking hardware and software to enable device- and application-level security, live patching of switches, and enhanced protection for Kubernetes and AI workloads. <a href="https://app.daily.dev/posts/VWEt8pf7l" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li>AI-generated software tests: the illusion of quality: Apparently, LLMs are now churning out software tests, and some folks are actually using them. The catch? These tests are often generated after the code, creating a ‘false sense of security’ because they haven’t been validated by actually failing. It’s like getting a participation trophy for testing, but the game was rigged from the start. Maybe we should make the AI write the code from the tests, instead of the other way around, as suggested by some unnamed critics. <a href="https://app.daily.dev/posts/wLM9iDorf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Cursor AI agents: building browsers ‘from scratch’ (with existing parts): Cursor’s CEO proudly announced their AI agents built a ‘fully functional web browser’ in a week, ‘from scratch,’ generating millions of lines of code. Turns out, independent developers found the browser barely compiles, doesn’t work, and relies heavily on existing projects like Servo and QuickJS. So, ‘from scratch’ now means ‘from a pile of other people’s code that we claim our AI wrote.’ Impressive marketing, less impressive engineering. <a href="https://app.daily.dev/posts/DovPZUpOP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI coding assistants: the silent killer of open source?: A research paper suggests that ‘vibe coding’ with AI agents, which assemble software from open-source repos without talking to maintainers, is leading to a ‘loss of traditional user feedback mechanisms.’ This, in turn, supposedly decreases maintainer incentives and will lead to a ‘decline in OSS quality and availability.’ So, AI is so good it’s making open source worse by being too efficient? Or maybe, just maybe, open source needs to figure out a sustainable funding model beyond hoping users file bug reports. <a href="https://app.daily.dev/posts/YRC8HFFhU" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>