Google Cloud will shut down T4 and P4 GPUs on August 1, 2027

Google Cloud posted end-of-support notices for three older NVIDIA GPUs in its Compute Engine release notes on September 22, 2026. T4 and P4 GPUs reach end of support on August 1, 2027. P100 GPUs already reached it on September 15, 2026, and were shut down.

What happens to the T4 and P4

Until August 1, 2027, T4 and P4 instances keep running normally. On that date, Google Cloud shuts down any that remain, and you can no longer create, start or access resources that use these GPUs. Commitments are already affected:

• You can no longer buy or renew 3-year committed use discounts (CUDs) for T4 or P4 GPUs.
• Existing 1-year and 3-year commitments stay in effect until they expire. If yours runs past the end date, Google asks you to contact your account team.

The T4 notice also covers Google services that use it. Colab Enterprise runtimes, GKE nodes, and models, jobs, endpoints and Workbench instances on Gemini Enterprise Agent Platform all need to move to a supported GPU.

P100 is already gone

The P100 page says that after September 15, 2026, you can’t create or access any Google Cloud resources that use NVIDIA P100 GPUs.

Where Google suggests you move

Google recommends two newer machine series:

Series	GPU	GPU memory	Google’s suggested uses
G2	NVIDIA L4	24 GB	AI inference, generative AI, video, light training
G4	NVIDIA RTX PRO 6000	96 GB	Simulation, 3D rendering, local LLM work, virtual workstations

Google says the L4 offers up to four times the inference performance of the T4. Check that the new GPU is available in your region and that you have quota for it before you migrate.

What to do

If you run inference or notebooks on T4s, plan the move well before August 2027, and don’t count on renewing a 3-year T4 commitment. Moving to G2 or G4 also changes your hourly price, so compare the cost before you switch.