Google Sets Daily Token Limits for BigQuery AI

The company has introduced automated consumption quotas to improve cost predictability for generative AI workflows.

Updated on Oct. 15, 2026 in Artificial Intelligence

Google Sets Daily Token Limits for BigQuery AI

Live Poll

Do you trust automated cloud usage limits to effectively help manage your organization's technology spending?

Google has made daily token quotas generally available for its BigQuery generative AI functions. These limits allow organizations to manage usage costs by capping token consumption at both the project and individual user levels.

Why it matters

By implementing these guardrails, businesses can better control expenses and prevent unexpected spending on AI-powered analytics. The quotas ensure that BigQuery AI workloads remain within predefined budget parameters.

Default project quotas are set at 200 billion input tokens and 20 billion output tokens daily. User-level limits are 150 billion for input and 15 billion for output, excluding embedding functions.

The players

Google

Google is a global technology company that provides cloud computing services and artificial intelligence tools for enterprise data management.

BigQuery

BigQuery is a serverless, multi-cloud data warehouse designed for business agility and large-scale data analysis.

The details

BigQuery will return a QuotaExceeded error and halt requests if a workload exceeds these defined limits. Enterprises can use the new AI.COUNT_TOKENS function to estimate consumption before running large-scale data queries.

November 2025: Google integrated AI-based SQL functions into BigQuery.
October 15, 2026: Daily token quotas became generally available for all users.

The Tech Race

This move adds mature financial controls to BigQuery's generative AI SQL functions, mirroring the evolution of cloud resource management. It positions Google to compete more effectively against rivals by offering enterprises greater predictability for AI-integrated data pipelines.

Developers and data engineers can now use the Cloud Quotas API or Terraform to enforce strict usage boundaries on their AI workloads. This helps teams avoid runaway costs by sampling smaller datasets before committing to full-scale AI processing.

The takeaway

These quotas provide a necessary safety net for teams integrating AI into their data architecture. Implementing sample workloads is now an essential step for teams aiming to optimize their AI usage under these new caps.