diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 4baecbc3b..b8c5dccc4 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -432,7 +432,7 @@ compile-training:
   extends: .rust-base-cpu
   stage: compile
   image: ${CI_BUILDER_IMAGE}
-  needs: [compile-services]  # sequential — CPU pool fits one 28-core job at a time
+  needs: []  # parallel — CPU pool autoscales to 2 nodes (POP2-32C-128G each)
   variables:
     # CUDA stubs from ci-builder image — compiles CUDA kernels without GPU hardware.
     # L4 GPU = compute capability 8.9 → candle-kernels generates sm_89 PTX