diff --git a/tensorflow/compiler/jit/BUILD b/tensorflow/compiler/jit/BUILD
index c844f6d1801..618165d4b64 100644
--- a/tensorflow/compiler/jit/BUILD
+++ b/tensorflow/compiler/jit/BUILD
@@ -115,6 +115,7 @@ cc_library(
     srcs = ["xla_gpu_device.cc"],
     visibility = [":friends"],
     deps = [
+        ":flags",
         ":jit_compilation_passes",
         ":xla_device",
         ":xla_kernel_creator",  # buildcleaner: keep
diff --git a/tensorflow/compiler/jit/flags.cc b/tensorflow/compiler/jit/flags.cc
index 1cf71298b05..991ad82daa1 100644
--- a/tensorflow/compiler/jit/flags.cc
+++ b/tensorflow/compiler/jit/flags.cc
@@ -155,6 +155,7 @@ void AllocateAndParseFlags() {
 
   device_flags = new XlaDeviceFlags;
   device_flags->tf_xla_compile_on_demand = false;
+  device_flags->tf_xla_enable_xla_devices = true;
 
   ops_flags = new XlaOpsCommonFlags;
   ops_flags->tf_xla_always_defer_compilation = false;
@@ -187,6 +188,12 @@ void AllocateAndParseFlags() {
             "Switch a device into 'on-demand' mode, where instead of "
             "autoclustering ops are compiled one by one just-in-time."),
 
+       Flag("tf_xla_enable_xla_devices",
+            &device_flags->tf_xla_enable_xla_devices,
+            "Generate XLA_* devices, where placing a computation on such a "
+            "device"
+            "forces compilation by XLA. Deprecated."),
+
        Flag("tf_xla_always_defer_compilation",
             &ops_flags->tf_xla_always_defer_compilation, ""),
 
diff --git a/tensorflow/compiler/jit/flags.h b/tensorflow/compiler/jit/flags.h
index 87a89841b91..618e839fa36 100644
--- a/tensorflow/compiler/jit/flags.h
+++ b/tensorflow/compiler/jit/flags.h
@@ -87,6 +87,9 @@ struct XlaDeviceFlags {
   // Enabling this mode by a legacy flag is a temporary mechanism. When this
   // feature is battle-tested, we will switch this to be a session option.
   bool tf_xla_compile_on_demand;
+
+  // Enables "XLA" devices if this flag is set.
+  bool tf_xla_enable_xla_devices;
 };
 
 // Flags common to the _Xla* ops and their kernels.
diff --git a/tensorflow/compiler/jit/xla_cpu_device.cc b/tensorflow/compiler/jit/xla_cpu_device.cc
index 85c09a027d3..446cd8944de 100644
--- a/tensorflow/compiler/jit/xla_cpu_device.cc
+++ b/tensorflow/compiler/jit/xla_cpu_device.cc
@@ -36,8 +36,13 @@ class XlaCpuDeviceFactory : public DeviceFactory {
 };
 
 Status XlaCpuDeviceFactory::ListPhysicalDevices(std::vector<string>* devices) {
-  devices->push_back(absl::StrCat("/physical_device:", DEVICE_XLA_CPU, ":0"));
+  XlaDeviceFlags* flags = GetXlaDeviceFlags();
+  if (!flags->tf_xla_enable_xla_devices) {
+    LOG(INFO) << "Not creating XLA devices, tf_xla_enable_xla_devices not set";
+    return Status::OK();
+  }
 
+  devices->push_back(absl::StrCat("/physical_device:", DEVICE_XLA_CPU, ":0"));
   return Status::OK();
 }
 
@@ -45,6 +50,10 @@ Status XlaCpuDeviceFactory::CreateDevices(
     const SessionOptions& session_options, const string& name_prefix,
     std::vector<std::unique_ptr<Device>>* devices) {
   XlaDeviceFlags* flags = GetXlaDeviceFlags();
+  if (!flags->tf_xla_enable_xla_devices) {
+    LOG(INFO) << "Not creating XLA devices, tf_xla_enable_xla_devices not set";
+    return Status::OK();
+  }
   bool compile_on_demand = flags->tf_xla_compile_on_demand;
 
   XlaOpRegistry::DeviceRegistration registration;
diff --git a/tensorflow/compiler/jit/xla_gpu_device.cc b/tensorflow/compiler/jit/xla_gpu_device.cc
index 8dc75c969a4..91943edd775 100644
--- a/tensorflow/compiler/jit/xla_gpu_device.cc
+++ b/tensorflow/compiler/jit/xla_gpu_device.cc
@@ -17,9 +17,11 @@ limitations under the License.
 // operators using XLA via the XLA "CUDA" (GPU) backend.
 
 #include <set>
+
 #include "absl/memory/memory.h"
 #include "absl/strings/numbers.h"
 #include "absl/strings/str_split.h"
+#include "tensorflow/compiler/jit/flags.h"
 #include "tensorflow/compiler/jit/kernels/xla_ops.h"
 #include "tensorflow/compiler/jit/xla_device.h"
 #include "tensorflow/compiler/jit/xla_device_ops.h"
@@ -61,6 +63,12 @@ class XlaGpuDeviceFactory : public DeviceFactory {
 };
 
 Status XlaGpuDeviceFactory::ListPhysicalDevices(std::vector<string>* devices) {
+  XlaDeviceFlags* flags = GetXlaDeviceFlags();
+  if (!flags->tf_xla_enable_xla_devices) {
+    LOG(INFO) << "Not creating XLA devices, tf_xla_enable_xla_devices not set";
+    return Status::OK();
+  }
+
   auto platform = se::MultiPlatformManager::PlatformWithName("CUDA");
   if (!platform.ok()) {
     // Treat failures as non-fatal; there might not be a GPU in the machine.
@@ -84,6 +92,12 @@ Status XlaGpuDeviceFactory::ListPhysicalDevices(std::vector<string>* devices) {
 Status XlaGpuDeviceFactory::CreateDevices(
     const SessionOptions& session_options, const string& name_prefix,
     std::vector<std::unique_ptr<Device>>* devices) {
+  XlaDeviceFlags* flags = GetXlaDeviceFlags();
+  if (!flags->tf_xla_enable_xla_devices) {
+    LOG(INFO) << "Not creating XLA devices, tf_xla_enable_xla_devices not set";
+    return Status::OK();
+  }
+
   XlaOpRegistry::DeviceRegistration registration;
   registration.compilation_device_name = DEVICE_GPU_XLA_JIT;
   registration.autoclustering_policy =
diff --git a/tensorflow/compiler/tf2xla/xla_op_registry.cc b/tensorflow/compiler/tf2xla/xla_op_registry.cc
index a43608bd434..b16dd3086fe 100644
--- a/tensorflow/compiler/tf2xla/xla_op_registry.cc
+++ b/tensorflow/compiler/tf2xla/xla_op_registry.cc
@@ -140,7 +140,7 @@ XlaOpRegistry::~XlaOpRegistry() = default;
 
   // Lazily register the CPU and GPU JIT devices the first time
   // GetCompilationDevice is called.
-  static void* registration_init = [&registry]() {
+  {
     MarkForCompilationPassFlags* flags = GetMarkForCompilationPassFlags();
     bool cpu_global_jit = flags->tf_xla_cpu_global_jit;
     VLOG(2) << "tf_xla_cpu_global_jit = " << cpu_global_jit;
@@ -162,9 +162,7 @@ XlaOpRegistry::~XlaOpRegistry() = default;
       registration.autoclustering_policy =
           XlaOpRegistry::AutoclusteringPolicy::kIfEnabledGlobally;
     }
-    return nullptr;
-  }();
-  (void)registration_init;
+  }
 
   mutex_lock lock(registry.mutex_);
   auto it = registry.compilation_devices_.find(device_name);