April 17, 2026
Blink and it boots
Show HN: Smol machines – subsecond coldstart, portable virtual machines
Tiny VMs boot in a blink; commenters clash over the 'Docker killer' claim
TLDR: A new tool promises tiny “computer-in-a-file” apps that start in under a second with strong isolation. The community is hyped but split, arguing over the bold “Docker killer” claim, comparing it to Firecracker, and demanding proof of the speed and features like GPU support
Hacker News just met its latest fixation: a tiny tool that runs your apps inside their own mini-computer (a virtual machine) and supposedly starts in under a second. It works on Mac and Linux, packs everything into a single portable file you can carry around, and locks the internet by default so unknown code can’t secretly phone home. The ex-AWS maker boldly calls it a “replacement for Docker,” saying containers add unnecessary bloat — cue instant “Docker killer” memes and side-eyes. Fans cheered the clean comparison chart and the promise of fast starts, elastic memory, and the ability to use your Git without exposing your secret keys. Skeptics swarmed the thread asking for proof and details. Check it out at smolmachines.com.
Then came the drama. One commenter shrugged, “neat sounds like Firecracker,” comparing it to Amazon’s server tech, while the creator flexed differences aimed at laptops and simple shipping. The peanut gallery wanted receipts: How is subsecond startup actually done? Can it use a GPU? Can you chain one machine to feed another that stays offline? The vibe: excited but cautious. The hype squad says “try it,” the security crowd loves the isolation, and the skeptics demand benchmarks. The running joke? Blink and it’s already booted — and the comments are even faster, with hot takes flying as fast as the blink-and-it-boots tagline
Key Points
- •smolvm is a CLI to run isolated per‑workload Linux VMs locally with sub‑second (<200 ms) cold starts on macOS and Linux.
- •Workloads can be packed into a single .smolmachine file for portable, dependency‑free execution on matching host architectures.
- •Networking is disabled by default; granular egress is supported via host allow‑lists, enhancing sandboxing for untrusted code.
- •The system runs on Hypervisor.framework (macOS) or KVM (Linux) using libkrun and a custom kernel (libkrunfw), with elastic memory via virtio balloon.
- •A comparison table positions smolvm against containers, Colima, QEMU, Firecracker, and Kata in terms of isolation model, boot times, and macOS support.