Vercel makes Elastic Builds GA
Vercel announced Elastic Build Machines are generally available, a move it says will cut costs and speed builds for teams relying on large, infra‑heavy projects. The release highlights improved build performance as a developer‑experience win for engineering teams. (x.com)
When Vercel builds an app, it spins up a machine to install code dependencies and package the site before deployment. On April 16, 2026, the company said its Elastic Build Machines are now generally available. (vercel.com) Elastic Build Machines let Vercel choose the size of that machine for each project instead of forcing teams to pick one fixed tier. Vercel said the feature can be set at the team or project level and scales between 4 and 30 virtual central processing units and 8 to 60 gigabytes of memory. (vercel.com) The company first opened Elastic Build Machines in beta on March 24, 2026, for all paid plans. In that launch note, Vercel said smaller projects could stay on Standard machines while heavier workloads could scale up automatically to Enhanced or Turbo machines. (vercel.com) Build speed has become a selling point for Vercel as teams ship larger Next.js apps, add more dependencies, and run more frequent preview deployments from Git. Vercel’s build docs say larger machines are aimed at apps with slow builds or projects that run out of resources during packaging. (vercel.com) The economics are part of the pitch. Vercel says Elastic usage is billed by central processing unit minute starting at $0.0035, while Standard machines do not add build usage charges unless teams use on-demand concurrency. (vercel.com) The release also fits into a broader rebuild of Vercel’s own infrastructure. In an October 30, 2024 engineering post, the company said its internal build platform, called Hive, had powered builds since November 2023 and helped deliver a 30% improvement in build performance. (vercel.com) Vercel has been moving customers toward bigger default build capacity even before this launch. Its Pro plan docs say new Pro projects now use Turbo build machines by default, with 30 virtual central processing units and 60 gigabytes of memory, alongside on-demand concurrent builds. (vercel.com) That leaves Elastic as a middle path between fixed machine sizes and manual tuning. For teams with uneven workloads, Vercel is betting that automatic sizing will cut both idle overprovisioning and the long waits that slow down every deploy. (vercel.com)