### Issue type
Bug
### Have you reproduced the bug with TensorFlow Nightly?
Y…es
### Source
binary
### TensorFlow version
tf 2.20.0, 2.21.0, 2.22.0dev20260530
### Custom code
No
### OS platform and distribution
AlmaLinux 9.8
### Mobile device
_No response_
### Python version
3.13.5
### Bazel version
_No response_
### GCC/compiler version
_No response_
### CUDA/cuDNN version
12/9.23.1.3
### GPU model and memory
_No response_
### Current behavior?
When installing from PyPI I expect that `tf.test.is_built_with_cuda()` to be true. The wheels doesn't seem to have been built with GPU support?
### Standalone code to reproduce the issue
```shell
python -m venv tf.env
source tf.env/bin/activate
pip install --upgrade pip
pip install "tf-nightly[and-cuda]"
python -c "import tensorflow as tf; print(tf.test.is_built_with_cuda())"
```
### Relevant log output
```shell
[vikren@n502 ~]$ uname -a
Linux n502 5.14.0-687.12.1.el9_8.aarch64+64k #1 SMP PREEMPT_DYNAMIC Sun Jun 7 13:46:46 EDT 2026 aarch64 aarch64 aarch64 GNU/Linux
[vikren@n502 ~]$ nvidia-smi
Sat Jun 13 10:53:52 2026
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 580.159.04 Driver Version: 580.159.04 CUDA Version: 13.0 |
+-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA GH200 120GB On | 00000019:01:00.0 Off | 0 |
| N/A 35C P0 92W / 900W | 0MiB / 97871MiB | 0% Default |
| | | Disabled |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| No running processes found |
+-----------------------------------------------------------------------------------------+
[vikren@n502 ~]$ python --version
Python 3.13.5
[vikren@n502 ~]$ python -m venv tf.env
[vikren@n502 ~]$ source tf.env/bin/activate
(tf.env) [vikren@n502 ~]$ pip install --upgrade pip
Requirement already satisfied: pip in ./tf.env/lib/python3.13/site-packages (25.1.1)
Collecting pip
Using cached pip-26.1.2-py3-none-any.whl.metadata (4.6 kB)
Using cached pip-26.1.2-py3-none-any.whl (1.8 MB)
Installing collected packages: pip
Attempting uninstall: pip
Found existing installation: pip 25.1.1
Uninstalling pip-25.1.1:
Successfully uninstalled pip-25.1.1
Successfully installed pip-26.1.2
(tf.env) [vikren@n502 ~]$ pip install "tf-nightly[and-cuda]==2.22.0.dev20260530"
Collecting tf-nightly==2.22.0.dev20260530 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached tf_nightly-2.22.0.dev20260530-cp313-cp313-manylinux_2_27_aarch64.whl.metadata (4.5 kB)
Collecting absl-py>=1.0.0 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached absl_py-2.4.0-py3-none-any.whl.metadata (3.3 kB)
Collecting astunparse>=1.6.0 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached astunparse-1.6.3-py2.py3-none-any.whl.metadata (4.4 kB)
Collecting flatbuffers>=25.9.23 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached flatbuffers-25.12.19-py2.py3-none-any.whl.metadata (1.0 kB)
Collecting gast!=0.5.0,!=0.5.1,!=0.5.2,>=0.2.1 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached gast-0.7.0-py3-none-any.whl.metadata (1.5 kB)
Collecting google_pasta>=0.1.1 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached google_pasta-0.2.0-py3-none-any.whl.metadata (814 bytes)
Collecting libclang>=13.0.0 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached libclang-18.1.1-py2.py3-none-manylinux2014_aarch64.whl.metadata (5.2 kB)
Collecting opt_einsum>=2.3.2 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached opt_einsum-3.4.0-py3-none-any.whl.metadata (6.3 kB)
Collecting packaging (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached packaging-26.2-py3-none-any.whl.metadata (3.5 kB)
Collecting protobuf<8.0.0,>=6.31.1 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached protobuf-7.35.1-cp310-abi3-manylinux2014_aarch64.whl.metadata (595 bytes)
Collecting requests<3,>=2.21.0 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached requests-2.34.2-py3-none-any.whl.metadata (4.8 kB)
Collecting setuptools (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached setuptools-82.0.1-py3-none-any.whl.metadata (6.5 kB)
Collecting six>=1.12.0 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached six-1.17.0-py2.py3-none-any.whl.metadata (1.7 kB)
Collecting termcolor>=1.1.0 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached termcolor-3.3.0-py3-none-any.whl.metadata (6.5 kB)
Collecting typing_extensions>=3.6.6 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached typing_extensions-4.15.0-py3-none-any.whl.metadata (3.3 kB)
Collecting wrapt>=1.11.0 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached wrapt-2.2.1-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl.metadata (7.4 kB)
Collecting grpcio<2.0,>=1.24.3 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached grpcio-1.81.1-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl.metadata (3.7 kB)
Collecting keras-nightly>=3.12.0.dev (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached keras_nightly-3.15.0.dev2026061305-py3-none-any.whl.metadata (6.3 kB)
Collecting numpy>=1.26.0 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached numpy-2.4.6-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl.metadata (6.6 kB)
Collecting h5py<3.15.0,>=3.11.0 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached h5py-3.14.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl.metadata (2.7 kB)
Collecting ml_dtypes<1.0.0,>=0.5.1 (from tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached ml_dtypes-0.5.4-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl.metadata (8.9 kB)
Collecting nvidia-cublas-cu12<13.0,>=12.5.3.2 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_cublas_cu12-12.9.2.10-py3-none-manylinux_2_27_aarch64.whl.metadata (1.8 kB)
Collecting nvidia-cuda-cupti-cu12<13.0,>=12.5.82 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_cuda_cupti_cu12-12.9.79-py3-none-manylinux_2_25_aarch64.whl.metadata (1.8 kB)
Collecting nvidia-cuda-nvcc-cu12<13.0,>=12.5.82 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_cuda_nvcc_cu12-12.9.86-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl.metadata (1.7 kB)
Collecting nvidia-cuda-nvrtc-cu12<13.0,>=12.5.82 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_cuda_nvrtc_cu12-12.9.86-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl.metadata (1.7 kB)
Collecting nvidia-cuda-runtime-cu12<13.0,>=12.5.82 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_cuda_runtime_cu12-12.9.79-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl.metadata (1.7 kB)
Collecting nvidia-cudnn-cu12<10.0,>=9.3.0.75 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_cudnn_cu12-9.23.1.3-py3-none-manylinux_2_27_aarch64.whl.metadata (1.9 kB)
Collecting nvidia-cufft-cu12<12.0,>=11.2.3.61 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_cufft_cu12-11.4.1.4-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl.metadata (1.8 kB)
Collecting nvidia-curand-cu12<11.0,>=10.3.6.82 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_curand_cu12-10.3.10.19-py3-none-manylinux_2_27_aarch64.whl.metadata (1.7 kB)
Collecting nvidia-cusolver-cu12<12.0,>=11.6.3.83 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_cusolver_cu12-11.7.5.82-py3-none-manylinux_2_27_aarch64.whl.metadata (1.9 kB)
Collecting nvidia-cusparse-cu12<13.0,>=12.5.1.3 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_cusparse_cu12-12.5.10.65-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl.metadata (1.8 kB)
Collecting nvidia-nccl-cu12<3.0,>=2.29.7 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_nccl_cu12-2.30.7-py3-none-manylinux_2_18_aarch64.whl.metadata (2.1 kB)
Collecting nvidia-nvjitlink-cu12<13.0,>=12.5.82 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_nvjitlink_cu12-12.9.86-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl.metadata (1.7 kB)
Collecting nvidia-nvshmem-cu12>=3.2.5 (from tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_nvshmem_cu12-3.7.0-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl.metadata (2.2 kB)
Collecting charset_normalizer<4,>=2 (from requests<3,>=2.21.0->tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached charset_normalizer-3.4.7-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl.metadata (40 kB)
Collecting idna<4,>=2.5 (from requests<3,>=2.21.0->tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached idna-3.18-py3-none-any.whl.metadata (6.1 kB)
Collecting urllib3<3,>=1.26 (from requests<3,>=2.21.0->tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached urllib3-2.7.0-py3-none-any.whl.metadata (6.9 kB)
Collecting certifi>=2023.5.7 (from requests<3,>=2.21.0->tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached certifi-2026.5.20-py3-none-any.whl.metadata (2.5 kB)
Collecting wheel<1.0,>=0.23.0 (from astunparse>=1.6.0->tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached wheel-0.47.0-py3-none-any.whl.metadata (2.3 kB)
Collecting rich (from keras-nightly>=3.12.0.dev->tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached rich-15.0.0-py3-none-any.whl.metadata (18 kB)
Collecting namex (from keras-nightly>=3.12.0.dev->tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached namex-0.1.0-py3-none-any.whl.metadata (322 bytes)
Collecting optree (from keras-nightly>=3.12.0.dev->tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached optree-0.19.1-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl.metadata (32 kB)
Collecting nvidia-cuda-cccl-cu12 (from nvidia-nvshmem-cu12>=3.2.5->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached nvidia_cuda_cccl_cu12-12.9.27-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl.metadata (1.7 kB)
Collecting markdown-it-py>=2.2.0 (from rich->keras-nightly>=3.12.0.dev->tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached markdown_it_py-4.2.0-py3-none-any.whl.metadata (7.4 kB)
Collecting pygments<3.0.0,>=2.13.0 (from rich->keras-nightly>=3.12.0.dev->tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached pygments-2.20.0-py3-none-any.whl.metadata (2.5 kB)
Collecting mdurl~=0.1 (from markdown-it-py>=2.2.0->rich->keras-nightly>=3.12.0.dev->tf-nightly==2.22.0.dev20260530->tf-nightly[and-cuda]==2.22.0.dev20260530)
Using cached mdurl-0.1.2-py3-none-any.whl.metadata (1.6 kB)
Using cached tf_nightly-2.22.0.dev20260530-cp313-cp313-manylinux_2_27_aarch64.whl (299.6 MB)
Using cached grpcio-1.81.1-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl (6.6 MB)
Using cached h5py-3.14.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (4.7 MB)
Using cached ml_dtypes-0.5.4-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl (5.0 MB)
Using cached nvidia_cublas_cu12-12.9.2.10-py3-none-manylinux_2_27_aarch64.whl (575.0 MB)
Using cached nvidia_cuda_cupti_cu12-12.9.79-py3-none-manylinux_2_25_aarch64.whl (10.1 MB)
Using cached nvidia_cuda_nvcc_cu12-12.9.86-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl (39.4 MB)
Using cached nvidia_cuda_nvrtc_cu12-12.9.86-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl (44.5 MB)
Using cached nvidia_cuda_runtime_cu12-12.9.79-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl (3.5 MB)
Using cached nvidia_cudnn_cu12-9.23.1.3-py3-none-manylinux_2_27_aarch64.whl (778.2 MB)
Using cached nvidia_cufft_cu12-11.4.1.4-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl (200.8 MB)
Using cached nvidia_curand_cu12-10.3.10.19-py3-none-manylinux_2_27_aarch64.whl (68.3 MB)
Using cached nvidia_cusolver_cu12-11.7.5.82-py3-none-manylinux_2_27_aarch64.whl (337.9 MB)
Using cached nvidia_cusparse_cu12-12.5.10.65-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl (366.4 MB)
Using cached nvidia_nccl_cu12-2.30.7-py3-none-manylinux_2_18_aarch64.whl (303.3 MB)
Using cached nvidia_nvjitlink_cu12-12.9.86-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl (39.5 MB)
Using cached protobuf-7.35.1-cp310-abi3-manylinux2014_aarch64.whl (328 kB)
Using cached requests-2.34.2-py3-none-any.whl (73 kB)
Using cached charset_normalizer-3.4.7-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl (207 kB)
Using cached idna-3.18-py3-none-any.whl (65 kB)
Using cached typing_extensions-4.15.0-py3-none-any.whl (44 kB)
Using cached urllib3-2.7.0-py3-none-any.whl (131 kB)
Using cached absl_py-2.4.0-py3-none-any.whl (135 kB)
Using cached astunparse-1.6.3-py2.py3-none-any.whl (12 kB)
Using cached six-1.17.0-py2.py3-none-any.whl (11 kB)
Using cached wheel-0.47.0-py3-none-any.whl (32 kB)
Using cached certifi-2026.5.20-py3-none-any.whl (134 kB)
Using cached flatbuffers-25.12.19-py2.py3-none-any.whl (26 kB)
Using cached gast-0.7.0-py3-none-any.whl (22 kB)
Using cached google_pasta-0.2.0-py3-none-any.whl (57 kB)
Using cached keras_nightly-3.15.0.dev2026061305-py3-none-any.whl (1.7 MB)
Using cached libclang-18.1.1-py2.py3-none-manylinux2014_aarch64.whl (23.8 MB)
Using cached numpy-2.4.6-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl (15.7 MB)
Using cached nvidia_nvshmem_cu12-3.7.0-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl (230.1 MB)
Using cached opt_einsum-3.4.0-py3-none-any.whl (71 kB)
Using cached packaging-26.2-py3-none-any.whl (100 kB)
Using cached termcolor-3.3.0-py3-none-any.whl (7.7 kB)
Using cached wrapt-2.2.1-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl (166 kB)
Using cached namex-0.1.0-py3-none-any.whl (5.9 kB)
Using cached nvidia_cuda_cccl_cu12-12.9.27-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl (3.2 MB)
Using cached optree-0.19.1-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl (417 kB)
Using cached rich-15.0.0-py3-none-any.whl (310 kB)
Using cached pygments-2.20.0-py3-none-any.whl (1.2 MB)
Using cached markdown_it_py-4.2.0-py3-none-any.whl (91 kB)
Using cached mdurl-0.1.2-py3-none-any.whl (10.0 kB)
Using cached setuptools-82.0.1-py3-none-any.whl (1.0 MB)
Installing collected packages: namex, libclang, flatbuffers, wrapt, urllib3, typing_extensions, termcolor, six, setuptools, pygments, protobuf, packaging, opt_einsum, nvidia-nvjitlink-cu12, nvidia-nccl-cu12, nvidia-curand-cu12, nvidia-cuda-runtime-cu12, nvidia-cuda-nvrtc-cu12, nvidia-cuda-nvcc-cu12, nvidia-cuda-cupti-cu12, nvidia-cuda-cccl-cu12, numpy, mdurl, idna, gast, charset_normalizer, certifi, absl-py, wheel, requests, optree, nvidia-nvshmem-cu12, nvidia-cusparse-cu12, nvidia-cufft-cu12, nvidia-cublas-cu12, ml_dtypes, markdown-it-py, h5py, grpcio, google_pasta, rich, nvidia-cusolver-cu12, nvidia-cudnn-cu12, astunparse, keras-nightly, tf-nightly
Successfully installed absl-py-2.4.0 astunparse-1.6.3 certifi-2026.5.20 charset_normalizer-3.4.7 flatbuffers-25.12.19 gast-0.7.0 google_pasta-0.2.0 grpcio-1.81.1 h5py-3.14.0 idna-3.18 keras-nightly-3.15.0.dev2026061305 libclang-18.1.1 markdown-it-py-4.2.0 mdurl-0.1.2 ml_dtypes-0.5.4 namex-0.1.0 numpy-2.4.6 nvidia-cublas-cu12-12.9.2.10 nvidia-cuda-cccl-cu12-12.9.27 nvidia-cuda-cupti-cu12-12.9.79 nvidia-cuda-nvcc-cu12-12.9.86 nvidia-cuda-nvrtc-cu12-12.9.86 nvidia-cuda-runtime-cu12-12.9.79 nvidia-cudnn-cu12-9.23.1.3 nvidia-cufft-cu12-11.4.1.4 nvidia-curand-cu12-10.3.10.19 nvidia-cusolver-cu12-11.7.5.82 nvidia-cusparse-cu12-12.5.10.65 nvidia-nccl-cu12-2.30.7 nvidia-nvjitlink-cu12-12.9.86 nvidia-nvshmem-cu12-3.7.0 opt_einsum-3.4.0 optree-0.19.1 packaging-26.2 protobuf-7.35.1 pygments-2.20.0 requests-2.34.2 rich-15.0.0 setuptools-82.0.1 six-1.17.0 termcolor-3.3.0 tf-nightly-2.22.0.dev20260530 typing_extensions-4.15.0 urllib3-2.7.0 wheel-0.47.0 wrapt-2.2.1
(tf.env) [vikren@n502 ~]$ python -c "import tensorflow as tf; print(tf.test.is_built_with_cuda())"
False
```