BigQuery Restores Daily Token Quotas for Generative AI Functions

BigQuery has restored daily token quotas for generative AI functions, giving teams a direct way to limit token consumption.

These quotas can help reduce the risk of unexpected generative artificial intelligence spend by placing a daily limit on usage. That gives FinOps and data teams a practical control for managing workloads that use BigQuery’s generative AI functions.

BigQuery Graph Billing Requirements Give Teams Time to Plan

Google Cloud has announced future billing requirements for BigQuery Graph core processing. After April 26, 2027, BigQuery Graph core processing will require an Enterprise or Enterprise Plus edition reservation.

Existing allowlisted users can temporarily continue using the Standard edition or on-demand billing. Even so, organizations using BigQuery Graph should assess the future reservation and edition cost impact.

Looker Adds Token Usage Visibility for Conversational Analytics

Looker’s Conversational Analytics System Activity dashboard now shows token usage by user and conversation, along with the type of data agent involved.

These metrics provide visibility into artificial intelligence consumption patterns. Teams can use that information to support chargeback, detect unusual usage, and optimize workloads.

Database Center Adds Assessments for Safer Right-Sizing

Database Center now provides assessments that test recommended configuration changes before production rollout. The assessments clone Cloud SQL for PostgreSQL instances, benchmark recommended changes, and compare the results.

This approach supports evidence-based right-sizing and performance optimization. Teams can evaluate whether a recommendation improves the configuration without applying a potentially costly or ineffective change directly to production.

Managed Kafka Adds Flexible Broker Disk Capacity

Google Cloud Managed Service for Apache Kafka now lets customers configure broker disk capacity when creating a cluster, with the option to increase capacity later.

More flexible storage sizing can reduce overprovisioning. Instead of allocating excessive capacity at the start, teams can allow storage to track workload growth.