diff --git a/vllm/config/compilation.py b/vllm/config/compilation.py
index 083b23aef92c..7a69629f707c 100644
--- a/vllm/config/compilation.py
+++ b/vllm/config/compilation.py
@@ -595,6 +595,24 @@ class CompilationConfig:
     local_cache_dir: str = field(default=None, init=False)  # type: ignore
     """local cache dir for each rank"""
 
+    fast_moe_cold_start = True
+    """Optimization for fast MOE cold start.
+
+    This is a bit of a hack that assumes that:
+    1. the only decoder forward pass being run is the current model
+    2. the decoder forward pass runs all of the MOEs in the order in which they
+       are initialized
+
+    When the above two conditions hold, this option greatly decreases cold start
+    time for MOE models.
+
+    If the above two conditions don't hold, then this option will lead to silent
+    incorrectness. The only condition in which this doesn't hold is speculative
+    decoding, where there is a draft model that may have MOEs in them.
+
+    NB: We're working on a longer-term solution that doesn't need these assumptions.
+    """
+
     # keep track of enabled and disabled custom ops
     enabled_custom_ops: Counter[str] = field(default_factory=Counter, init=False)
     """custom ops that are enabled"""
diff --git a/vllm/forward_context.py b/vllm/forward_context.py
index a5c833b5e433..e308c05bc669 100644
--- a/vllm/forward_context.py
+++ b/vllm/forward_context.py
@@ -286,9 +286,22 @@ def create_forward_context(
     additional_kwargs: dict[str, Any] | None = None,
     skip_compiled: bool = False,
 ):
+    if vllm_config.compilation_config.fast_moe_cold_start:
+        if vllm_config.speculative_config is None:
+            all_moe_layers = vllm_config.compilation_config.static_all_moe_layers
+        else:
+            logger.warning_once(
+                "vllm_config.compilation_config.fast_moe_cold_start is not "
+                "compatible with speculative decoding so we are ignoring "
+                "fast_moe_cold_start."
+            )
+            all_moe_layers = None
+    else:
+        all_moe_layers = None
+
     return ForwardContext(
         no_compile_layers=vllm_config.compilation_config.static_forward_context,
-        all_moe_layers=vllm_config.compilation_config.static_all_moe_layers,
+        all_moe_layers=all_moe_layers,
         virtual_engine=virtual_engine,
         attn_metadata=attn_metadata,
         slot_mapping=slot_mapping or {},