I ran 20 AI coding agents on one PC. The bottleneck was the compiler.

A developer ran 20 parallel AI coding agents on a single desktop PC and found the compiler, not the model, was the bottleneck: twenty concurrent git worktrees each triggered cold builds and test runs, exhausting RAM and CPU while the GPU sat idle. By hashing the source tree, toolchain and command to cache build results, 83–85% of checks were served from cache, and the setup completed 1,770 of 1,770 checks green across 4 rounds with 20 agents. The developer's takeaway is that adding agents only helps once verification is cheap and impossible to fake.

Everyone argues about which model writes the best code. When I ran 20 coding agents in parallel on one PC, the model was never what slowed things down. The compiler was. Parallel agents usually work in separate copies of the repo git worktrees . Twenty agents means twenty copies, and every one of them wants to build and run the test suite after each change. That's twenty cold builds at once. On a normal desktop, RAM runs out first, then the CPU, and the agents sit waiting on cargo test while the GPU running the model idles. Most of those builds are redundant. Agents working on the same task often land on identical code, and most files are untouched in any given change. So I hash the inputs the source tree, the toolchain, the command and cache the result. Same inputs, same answer, no rebuild. In my runs, 83 to 85% of checks were served from that cache. "Done" is not the agent saying it's done. Done is a real build and the real tests passing, run by something outside the agent. Source is writable. Tests and the grader are read-only. Otherwise a stuck agent will eventually "fix" the failing test instead of the code. It's not malicious, it's just the shortest path to green. I cap RAM 21 of 32 GB on my machine and throttle build jobs. A swarm that crashes your PC finishes nothing. Agents work in parallel, but changes land one by one. A change only goes in if more tests pass and none newly fail. Parallel work, serial truth. 20 agents, 4 rounds, 1,770 of 1,770 checks green. The takeaway: adding agents only helps once verification is cheap and impossible to fake. Until then, more agents just means more broken code, faster. What's the bottleneck in your setup?