·8 min read·

The Apple Silicon transition: the studio just got faster.

We've been running M1 MacBook Pros for three weeks. Builds are faster, fans are silent, battery actually means something. Here is what it changes for a small studio, the real benchmarks, the gotchas, and the buy-now recommendation per role.

Apple announced the M1 Macs in November. We bought two MacBook Pros to test them in real workloads — Nuxt builds, NativeScript Android packaging, multi-tab Chrome, a virtual machine running Windows for Visual Studio. Three weeks in, the verdict is striking. This is the first laptop generation in years that genuinely changes the shape of how a studio operates.

Real benchmarks, our workloads

Task2019 Intel MBP (i9, 32GB)M1 MBP (8GB)M1 MBP (16GB)
Nuxt cold build2m 14s1m 28s1m 22s
Webpack hot reload (typical)~3s<1s<1s
NativeScript Android packaging4m 12s2m 41s2m 35s
Single tab YouTube + Slack idleFans on, ~25WSilent, ~6WSilent, ~7W
10-tab dev session, fans engagedYes within 5minNever (no fans)Never (no fans)
Battery — typical mixed dev work~3 hours~10 hours~10 hours
Measured workload performance, December 2020. 2019 Intel reference is i9 / 32GB / Vega 20.

What got measurably faster

  • 01Nuxt build times dropped roughly 35% versus our 2019 Intel Pros.
  • 02Webpack hot reloads feel instant rather than 'noticeable'.
  • 03Multi-tab Chrome with eight Cursor / video / Slack / dev servers — no fan, no thermal throttle.
  • 04Battery life is approaching what Apple claims. That hasn't been true on a Mac laptop for years.
  • 05Wake from sleep is instant. The MacBook is now a phone-shaped experience.
  • 06Compile times for native Mac apps (Xcode projects) drop dramatically.

What broke or wobbled

  • 01Some Node modules with native bindings needed Rosetta. Most have ARM builds within weeks now.
  • 02Docker is slower than on Intel — it's running Linux under emulation. Improving, not fast yet.
  • 03A few production Electron apps had rough edges on launch day. All since patched.
  • 04Some Homebrew formulas need manual ARM builds. The bottle situation has improved fast but is not yet complete.
  • 05x86-only commercial software (some Adobe utilities, some VPN clients) runs under Rosetta and is mostly fine but occasionally weird.

Hardware rarely changes the shape of a studio. This one did. The build server we were about to buy is suddenly unnecessary — three M1 laptops together rival what an Intel Mac mini cluster gave us six months ago.

What we'd buy now (December 2020)

RoleRecommended configWhy
Engineer (web)M1 MacBook Pro 13, 16GB, 512GBFuture-proof for two years; 16GB matters for Docker
Engineer (native / heavy)Wait for M1 Pro (rumoured 2021)Base M1 maxes at 16GB; pro workloads will want more
Designer (Figma)M1 MacBook Air, 8GB or 16GBGenuinely fast enough; cheaper, lighter, no fan needed
Anyone running heavy x86 Docker / VMsKeep Intel for nowRosetta is fast but VMs are slow
Installations / kiosksM1 Mac miniCheap, silent, fanless, perfect for permanent installs
Apple Silicon buying matrix, December 2020.

Hardware rarely changes the shape of a studio. This one did. The build server we were about to buy is suddenly unnecessary. Engineers can take their MacBook to a coffee shop and have it last a working day. The whole thermal story has flipped — no fans, no throttling, full performance at room temperature. The M1 isn't an iteration. It's a different category of laptop, and Apple have brought it to market two years before anyone else.

Talk to Remiam about a system like this.