diff --git a/tensorflow/compiler/xla/pjrt/tpu_client.cc b/tensorflow/compiler/xla/pjrt/tpu_client.cc
index 830f7c66f2a..95f581479e5 100644
--- a/tensorflow/compiler/xla/pjrt/tpu_client.cc
+++ b/tensorflow/compiler/xla/pjrt/tpu_client.cc
@@ -68,29 +68,8 @@ Status TpuDeviceState::ThenMemcpyDeviceToDevice(
     se::DeviceMemoryBase src_buffer, se::DeviceMemoryBase dst_buffer) {
   auto* transfer_tpu_stream = tensorflow::down_cast<tf_tpu::TpuStream*>(
       transfer_stream->implementation());
-  tf_tpu::TpuTopologyExternal topology =
-      tf_tpu::TpuPlatformInterface::GetRegisteredPlatform()->topology();
-  // TODO(b/157179600): use device-to-device transfers when implemented instead
-  // of copying via host.
-  if (topology.version() == kTpuV4) {
-    LOG(WARNING)
-        << "device-to-device transfers not yet implemented, copying via host";
-    auto* dst_tpu_stream =
-        tensorflow::down_cast<tf_tpu::TpuStream*>(dst_stream->implementation());
-    TF_RET_CHECK(src_buffer.size() == dst_buffer.size());
-    auto host_tmp = std::make_unique<char[]>(src_buffer.size());
-    TF_RETURN_IF_ERROR(transfer_tpu_stream->EnqueueTransferDeviceToHost(
-        src_buffer, host_tmp.get(), src_buffer.size()));
-    dst_stream->ThenWaitFor(transfer_stream);
-    TF_RETURN_IF_ERROR(dst_tpu_stream->EnqueueTransferHostToDevice(
-        dst_buffer, host_tmp.get(), dst_buffer.size()));
-    transfer_stream->ThenWaitFor(dst_stream);
-    char* tmp = host_tmp.release();
-    dst_stream->ThenDoHostCallback([tmp] { delete[] tmp; });
-  } else {
-    TF_RETURN_IF_ERROR(transfer_tpu_stream->EnqueueOnTpuDeviceSendRecvLocal(
-        src_buffer, dst_buffer));
-  }
+  TF_RETURN_IF_ERROR(transfer_tpu_stream->EnqueueOnTpuDeviceSendRecvLocal(
+      src_buffer, dst_buffer));
   return Status::OK();
 }