Google Cloud released a comprehensive set of infrastructure updates on September 1, 2026, introducing new AI capabilities across multiple services, though notably absent from the announcements is any mention of the Veo video generation model. The release notes reveal expansions in generative AI functions, image generation, and enterprise compliance features, but do not address video generation tools or DeepMind's Veo specifically.

What New AI Capabilities Did Google Cloud Actually Announce?

Google Cloud's latest release notes highlight several significant AI and data processing updates. The platform now supports TabFM, a pre-trained foundation model for tabular data that enables zero-shot regression and classification through in-context learning. This means users can get high-accuracy predictions on structured data without requiring model training or hyperparameter tuning. The model is available through the AI.PREDICT function and can be evaluated using the AI.EVALUATE function, currently in preview.

On the image generation front, Google expanded capabilities for Gemini 3.1 Flash Image and Gemini 3 Pro Image. These models are now available across US and EU multi-region endpoints, and 4K image output has moved from preview to general availability. Video inputs for image generation are also now generally available, allowing users to generate images directly from video content.

Which Enterprise and Compliance Features Expanded?

Beyond AI model updates, Google Cloud introduced several enterprise-focused features. Gemini Enterprise now supports federated data stores in Assured Workloads with FedRAMP High compliance, allowing organizations to connect Google Workspace and third-party data stores while maintaining strict security controls. Gemini 3.5 Flash is now available in Canada with in-region data residency and machine learning processing. Additionally, Sensitive Data Protection content policies can now be applied to Gemini Enterprise connectors, apps, and Gemini Notebook Enterprise notebooks to block sensitive or inappropriate content.

Infrastructure improvements also include session affinity support for GKE Gateway load balancers, now generally available with multiple session affinity types including a new STRONG_COOKIE_AFFINITY option. BigQuery received several updates, including support for identity columns (auto-incrementing columns for primary keys), new graph query capabilities, and restored support for configuring daily token quotas for generative AI functions.

How to Navigate Google Cloud's Latest AI Infrastructure Changes

• TabFM for Data Analysis: Organizations working with structured data can now use TabFM directly through BigQuery without manual model training, reducing time and complexity for predictive analytics on tabular datasets.
• Image Generation at Scale: Teams needing 4K image generation or video-to-image conversion can access these capabilities through Gemini 3.1 Flash Image across multiple regions, enabling faster deployment across geographic locations.
• Compliance and Security: Enterprises in regulated industries can leverage FedRAMP High compliance support and Sensitive Data Protection policies to meet strict data handling requirements while using Gemini Enterprise features.
• Load Balancing Improvements: Development teams can implement session affinity for GKE Gateway load balancers, ensuring consistent user sessions across distributed infrastructure with multiple affinity type options.

What's Notably Missing From This Release?

Despite Google's known work on the Veo video generation model through DeepMind, the September 2026 release notes contain no announcements about video generation capabilities, Veo availability, or video processing infrastructure updates. The absence is notable given the competitive intensity in the video generation space, where companies like Alibaba and ByteDance have made significant public investments. Google's release focuses instead on tabular data models, image generation, and enterprise compliance features.

This silence could reflect several possibilities. Google may be developing Veo through separate channels not covered in standard cloud release notes, or the model may still be in research phases without production-ready infrastructure. Alternatively, Google may be prioritizing other AI capabilities for immediate cloud platform integration. Without explicit announcements, the status and timeline for Veo's enterprise availability remain unclear based on publicly available information.

For teams interested in Google's video generation capabilities, the current release notes suggest that image generation and video input processing are the primary video-related features available through Google Cloud's standard offerings. Organizations seeking dedicated video generation tools may need to monitor future announcements or explore alternative platforms currently offering production-ready video generation services.