diff --git a/onnxruntime/core/providers/openvino/openvino_provider_factory.cc b/onnxruntime/core/providers/openvino/openvino_provider_factory.cc
index cd82367b71a49..3929e83f56757 100644
--- a/onnxruntime/core/providers/openvino/openvino_provider_factory.cc
+++ b/onnxruntime/core/providers/openvino/openvino_provider_factory.cc
@@ -374,13 +374,12 @@ static void ParseProviderInfo(const ProviderOptions& provider_options,
   }
 
   // Should likely account for meta devices as well, but for now keep the current behavior.
-  bool target_devices_support_dynamic_shapes =
-      pi.device_type.find("GPU") != std::string::npos ||
-      pi.device_type.find("CPU") != std::string::npos ||
-      (pi.device_type.find("NPU") != std::string::npos &&
-       pi.enable_causallm);
-
-  pi.disable_dynamic_shapes = !target_devices_support_dynamic_shapes;
+  // Respect the user-provided option for CPU/GPU. For NPU, keep the existing constraint:
+  const bool is_npu_device = pi.device_type.find("NPU") != std::string::npos;
+  if (is_npu_device && pi.enable_causallm && pi.disable_dynamic_shapes) {
+    LOGS_DEFAULT(WARNING) << "Enabling dynamic shapes for NPU because CausalLM is enabled.";
+    pi.disable_dynamic_shapes = false;
+  }
 }
 
 struct OpenVINOProviderFactory : IExecutionProviderFactory {