diff --git a/.buildkite/test_areas/distributed.yaml b/.buildkite/test_areas/distributed.yaml
index 7875c5c511c2..5b27be0c38ef 100644
--- a/.buildkite/test_areas/distributed.yaml
+++ b/.buildkite/test_areas/distributed.yaml
@@ -4,6 +4,7 @@ depends_on:
 steps:
 - label: ":nvidia: (L4) Distributed Comm Ops"
   key: distributed-comm-ops
+  automatic_shard: true
   timeout_in_minutes: 25
   working_dir: "/vllm-workspace/tests"
   device: l4
@@ -21,6 +22,7 @@ steps:
   - pytest -v -s distributed/test_shm_storage.py
   mirror:
     amd:
+      automatic_shard: true
       label: ':amd: (MI300) Distributed Comm Ops'
       dind: false
       device: mi300_2
@@ -40,6 +42,7 @@ steps:
 
 - label: ":nvidia: (L4) Sharded RDT Weight Transfer"
   key: sharded-rdt-weight-transfer
+  automatic_shard: true
   device: l4
   timeout_in_minutes: 20
   working_dir: "/vllm-workspace/tests"
@@ -55,6 +58,7 @@ steps:
   - pytest -v -s distributed/test_sharded_rdt_trainer.py
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355 DPX) Sharded RDT Weight Transfer"
       dind: false
       device: mi355_dpx
@@ -130,6 +134,7 @@ steps:
 
 - label: ":nvidia: (L4) Distributed Compile + RPC"
   key: distributed-compile-rpc-tests-2-gpus
+  automatic_shard: true
   timeout_in_minutes: 65
   working_dir: "/vllm-workspace/tests"
   device: l4
@@ -152,6 +157,7 @@ steps:
   - pytest -v -s ./compile/test_wrapper.py
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) Distributed Compile + RPC"
       dind: false
       device: mi355_2
@@ -176,6 +182,7 @@ steps:
 
 - label: ":nvidia: (L4) Distributed Torchrun + Shutdown"
   key: distributed-torchrun-shutdown-tests-2-gpus
+  automatic_shard: true
   timeout_in_minutes: 30
   working_dir: "/vllm-workspace/tests"
   device: l4
@@ -200,6 +207,7 @@ steps:
   - pytest -v -s v1/worker/test_worker_memory_snapshot.py
   mirror:
     amd:
+      automatic_shard: true
       label: ':amd: (MI300) Distributed Torchrun + Shutdown'
       dind: false
       device: mi300_2
@@ -369,6 +377,7 @@ steps:
 
 - label: ":nvidia: (L4) [Nightly] Distributed Compile + Comm"
   key: distributed-compile-comm-4-gpus
+  automatic_shard: true
   # Nightly only: 4-GPU L4 pods wait behind smaller jobs on per-PR runs.
   optional: true
   timeout_in_minutes: 70
@@ -392,6 +401,7 @@ steps:
   - pytest -v -s distributed/test_multiproc_executor.py::test_multiproc_executor_multi_node
   mirror:
     amd:
+      automatic_shard: true
       # Run on PRs even when the L4 step is optional (nightly only).
       optional: false
       label: ':amd: (MI355) Distributed Compile + Comm'
@@ -416,6 +426,7 @@ steps:
 
 - label: ":nvidia: (H100) [Nightly] Distributed DP + EP"
   key: distributed-tests-8xh100
+  automatic_shard: true
   timeout_in_minutes: 20
   optional: true
   device: h100
@@ -441,6 +452,7 @@ steps:
   - pytest -v -s distributed/test_eplb_execute.py -k symm_mem
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) Distributed DP + EP"
       dind: false
       device: mi355_8
@@ -577,6 +589,7 @@ steps:
 
 - label: ":nvidia: (B200) [Nightly] Distributed"
   key: distributed-tests-2xb200
+  automatic_shard: true
   device: b200-k8s
   optional: true
   working_dir: "/vllm-workspace/"
@@ -589,6 +602,7 @@ steps:
     - pytest -v -s tests/distributed/test_mnnvl_alltoall.py
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) Distributed AgRs All2All"
       dind: false
       device: mi355_2
@@ -611,6 +625,7 @@ steps:
 
 - label: ":nvidia: (B200) GEMM-RS/AR"
   key: gemm-rs-ar-2xb200
+  automatic_shard: true
   timeout_in_minutes: 20
   device: b200-k8s
   working_dir: "/vllm-workspace/"
@@ -623,6 +638,7 @@ steps:
     - pytest -v -s tests/kernels/test_gemm_rs_ar.py
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) GEMM-RS-AR"
       dind: false
       device: mi355_4
@@ -779,6 +795,7 @@ steps:
 
 - label: ":nvidia: (L4) Distributed Preload"
   key: preload-2-gpus
+  automatic_shard: true
   timeout_in_minutes: 15
   working_dir: "/vllm-workspace/tests"
   device: l4
@@ -791,6 +808,7 @@ steps:
   - pytest -v -s model_executor/model_loader/test_weight_cache.py -m 'distributed(num_gpus=2)'
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) Distributed Preload"
       dind: false
       device: mi355_2
diff --git a/.buildkite/test_areas/expert_parallelism.yaml b/.buildkite/test_areas/expert_parallelism.yaml
index 1ca3eacd48cb..e9c13dad6348 100644
--- a/.buildkite/test_areas/expert_parallelism.yaml
+++ b/.buildkite/test_areas/expert_parallelism.yaml
@@ -4,6 +4,7 @@ depends_on:
 steps:
 - label: ":nvidia: (H200 MIG 18GB) EPLB Algorithm"
   key: eplb-algorithm
+  automatic_shard: true
   timeout_in_minutes: 20
   device: h200_18gb
   working_dir: "/vllm-workspace/tests"
@@ -18,6 +19,7 @@ steps:
   - pytest -v -s distributed/test_eplb_utils.py
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355 DPX) EPLB Algorithm"
       dind: false
       device: mi355_dpx
@@ -32,6 +34,7 @@ steps:
 
 - label: ":nvidia: (L4) [Nightly] EPLB Execution"
   key: eplb-execution
+  automatic_shard: true
   # Nightly only: 4-GPU L4 pods wait behind smaller jobs on per-PR runs.
   optional: true
   timeout_in_minutes: 25
@@ -46,6 +49,7 @@ steps:
   - pytest -v -s distributed/test_eplb_spec_decode.py
   mirror:
     amd:
+      automatic_shard: true
       # Run on PRs even when the L4 step is optional (nightly only).
       optional: false
       label: ":amd: (MI355) EPLB Execution"
diff --git a/.buildkite/test_areas/fault_tolerance.yaml b/.buildkite/test_areas/fault_tolerance.yaml
index f9439880d454..b68d9b0fb1e4 100644
--- a/.buildkite/test_areas/fault_tolerance.yaml
+++ b/.buildkite/test_areas/fault_tolerance.yaml
@@ -62,6 +62,7 @@ steps:
 # Keep the AMD job: its commands do not run this installation.
 - label: ':amd: (MI355) MoRI EP Healthy Baseline'
   key: amd-fault-tolerance-e2e-2xh100
+  automatic_shard: true
   timeout_in_minutes: 10
   device: mi355_2
   num_devices: 2
diff --git a/.buildkite/test_areas/models_distributed.yaml b/.buildkite/test_areas/models_distributed.yaml
index bdaef9ec5a1b..22e38ab1aa24 100644
--- a/.buildkite/test_areas/models_distributed.yaml
+++ b/.buildkite/test_areas/models_distributed.yaml
@@ -4,6 +4,7 @@ depends_on:
 steps:
 - label: ":nvidia: (L4) [Nightly] Qwen4 Exp SP"
   key: qwen4-exp-sp-tests-4-gpus
+  automatic_shard: true
   # Nightly only: 4-GPU L4 pods wait behind smaller jobs on per-PR runs.
   optional: true
   timeout_in_minutes: 30
@@ -22,6 +23,7 @@ steps:
   - pytest -v -s models/qwen4_exp/test_tpsp.py
   mirror:
     amd:
+      automatic_shard: true
       # Run on PRs even when the L4 step is optional (nightly only).
       optional: false
       label: ":amd: (MI355) Qwen4 Exp SP"