diff --git a/.github/workflows/slow.yml b/.github/workflows/slow.yml
index 097dcbdda..5231d9bb6 100644
--- a/.github/workflows/slow.yml
+++ b/.github/workflows/slow.yml
@@ -16,6 +16,7 @@ env:
   PYTHON_VERSION: "3.9"
   HATCH_VERSION: "1.14.1"
   HAYSTACK_MPS_ENABLED: false
+  HAYSTACK_XPU_ENABLED: false
 
 on:
   workflow_dispatch: # Activate this workflow manually
diff --git a/.github/workflows/tests.yml b/.github/workflows/tests.yml
index 46c02b5b5..bec43f08c 100644
--- a/.github/workflows/tests.yml
+++ b/.github/workflows/tests.yml
@@ -401,6 +401,9 @@ jobs:
     needs: unit-tests
     runs-on: windows-latest
     timeout-minutes: 30
+    env:
+      HAYSTACK_XPU_ENABLED: false
+
     steps:
       - uses: actions/checkout@v4
 
diff --git a/haystack/utils/device.py b/haystack/utils/device.py
index 256ca14eb..38653a421 100644
--- a/haystack/utils/device.py
+++ b/haystack/utils/device.py
@@ -28,6 +28,7 @@ class DeviceType(Enum):
     GPU = "cuda"
     DISK = "disk"
     MPS = "mps"
+    XPU = "xpu"
 
     def __str__(self):
         return self.value
@@ -126,6 +127,16 @@ class Device:
         """
         return Device(DeviceType.MPS)
 
+    @staticmethod
+    def xpu() -> "Device":
+        """
+        Create a generic Intel GPU Optimization device.
+
+        :returns:
+            The XPU device.
+        """
+        return Device(DeviceType.XPU)
+
     @staticmethod
     def from_str(string: str) -> "Device":
         """
@@ -482,7 +493,7 @@ def _get_default_device() -> Device:
     Return the default device for Haystack.
 
     Precedence:
-        GPU > MPS > CPU. If PyTorch is not installed, only CPU is available.
+        GPU > XPU > MPS > CPU. If PyTorch is not installed, only CPU is available.
 
     :returns:
         The default device.
@@ -496,12 +507,21 @@ def _get_default_device() -> Device:
             and os.getenv("HAYSTACK_MPS_ENABLED", "true") != "false"
         )
         has_cuda = torch.cuda.is_available()
+        has_xpu = (
+            hasattr(torch, "xpu")
+            and hasattr(torch.xpu, "is_available")
+            and torch.xpu.is_available()
+            and os.getenv("HAYSTACK_XPU_ENABLED", "true") != "false"
+        )
     except ImportError:
         has_mps = False
         has_cuda = False
+        has_xpu = False
 
     if has_cuda:
         return Device.gpu()
+    elif has_xpu:
+        return Device.xpu()
     elif has_mps:
         return Device.mps()
     else:
diff --git a/test/utils/test_device.py b/test/utils/test_device.py
index 88cd68d5c..997de6ee1 100644
--- a/test/utils/test_device.py
+++ b/test/utils/test_device.py
@@ -22,12 +22,14 @@ def test_device_creation():
     assert Device.cpu().type == DeviceType.CPU
     assert Device.gpu().type == DeviceType.GPU
     assert Device.mps().type == DeviceType.MPS
+    assert Device.xpu().type == DeviceType.XPU
     assert Device.disk().type == DeviceType.DISK
 
     assert Device.from_str("cpu") == Device.cpu()
     assert Device.from_str("cuda:1") == Device.gpu(1)
     assert Device.from_str("disk") == Device.disk()
     assert Device.from_str("mps:0") == Device(DeviceType.MPS, 0)
+    assert Device.from_str("xpu:0") == Device(DeviceType.XPU, 0)
 
     with pytest.raises(ValueError, match="Device id must be >= 0"):
         Device.gpu(-1)
@@ -115,23 +117,38 @@ def test_component_device_multiple():
     assert multiple.first_device == ComponentDevice.from_single(Device.cpu())
 
 
+@patch("torch.xpu.is_available")
 @patch("torch.backends.mps.is_available")
 @patch("torch.cuda.is_available")
-def test_component_device_resolution(torch_cuda_is_available, torch_backends_mps_is_available):
+def test_component_device_resolution(torch_cuda_is_available, torch_backends_mps_is_available, torch_xpu_is_available):
     assert ComponentDevice.resolve_device(ComponentDevice.from_single(Device.cpu()))._single_device == Device.cpu()
 
     torch_cuda_is_available.return_value = True
     assert ComponentDevice.resolve_device(None)._single_device == Device.gpu(0)
 
     torch_cuda_is_available.return_value = False
+    torch_xpu_is_available.return_value = True
+    torch_backends_mps_is_available.return_value = False
+    assert ComponentDevice.resolve_device(None)._single_device == Device.xpu()
+
+    torch_cuda_is_available.return_value = False
+    torch_xpu_is_available.return_value = False
     torch_backends_mps_is_available.return_value = True
     assert ComponentDevice.resolve_device(None)._single_device == Device.mps()
 
     torch_cuda_is_available.return_value = False
+    torch_xpu_is_available.return_value = False
     torch_backends_mps_is_available.return_value = False
     assert ComponentDevice.resolve_device(None)._single_device == Device.cpu()
 
     torch_cuda_is_available.return_value = False
+    torch_xpu_is_available.return_value = False
     torch_backends_mps_is_available.return_value = True
     os.environ["HAYSTACK_MPS_ENABLED"] = "false"
     assert ComponentDevice.resolve_device(None)._single_device == Device.cpu()
+
+    torch_cuda_is_available.return_value = False
+    torch_xpu_is_available.return_value = True
+    os.environ["HAYSTACK_XPU_ENABLED"] = "false"
+    torch_backends_mps_is_available.return_value = False
+    assert ComponentDevice.resolve_device(None)._single_device == Device.cpu()