diff --git a/jax/_src/hardware_utils.py b/jax/_src/hardware_utils.py
index 3f6686d05ee8..7fa1c0a5348c 100644
--- a/jax/_src/hardware_utils.py
+++ b/jax/_src/hardware_utils.py
@@ -36,6 +36,11 @@
     '0x005e',
 ]
 
+_NVIDIA_GPU_DEVICES = [
+    '/dev/nvidia0',
+    '/dev/dxg',  # WSL2
+]
+
 def num_available_tpu_chips_and_device_id():
   """Returns the device id and number of TPU chips attached through PCI."""
   num_chips = 0
@@ -57,3 +62,9 @@ def tpu_enhanced_barrier_supported() -> bool:
   """Returns if tpu_enhanced_barrier flag is supported on this TPU version."""
   _, device_id = num_available_tpu_chips_and_device_id()
   return device_id in _TPU_ENHANCED_BARRIER_SUPPORTED
+
+
+def has_visible_nvidia_gpu() -> bool:
+  """True if there's a visible nvidia gpu available on device, False otherwise."""
+
+  return any(os.path.exists(d) for d in _NVIDIA_GPU_DEVICES)
diff --git a/jax/_src/xla_bridge.py b/jax/_src/xla_bridge.py
index 243705490f81..3b2f11e21f17 100644
--- a/jax/_src/xla_bridge.py
+++ b/jax/_src/xla_bridge.py
@@ -881,6 +881,9 @@ def backends() -> dict[str, xla_client.Client]:
     default_priority = -1000
     for platform, priority, fail_quietly in platform_registrations:
       try:
+        if platform == "cuda" and not hardware_utils.has_visible_nvidia_gpu():
+          continue
+
         backend = _init_backend(platform)
         _backends[platform] = backend
 
@@ -918,12 +921,7 @@ def _suggest_missing_backends():
 
   assert _default_backend is not None
   default_platform = _default_backend.platform
-  nvidia_gpu_devices = [
-    "/dev/nvidia0",
-    "/dev/dxg",  # WSL2
-  ]
-  if ("cuda" not in _backends and
-      any(os.path.exists(d) for d in nvidia_gpu_devices)):
+  if "cuda" not in _backends and hardware_utils.has_visible_nvidia_gpu():
     if hasattr(xla_extension, "GpuAllocatorConfig") and "cuda" in _backend_errors:
       err = _backend_errors["cuda"]
       warning_msg = f"CUDA backend failed to initialize: {err}."