This week's briefing

What happened this week

This week’s defining pattern was AI moving deeper into operational systems—and making its constraints harder to ignore. From cross-Region model routing and agent-serving infrastructure to robotics, cyber operations, power, and financing, the biggest developments paired new capacity with explicit boundaries that carry into next week.

Lead story / launch

AWS Brings GPT-5.6 Cross-Region Inference, With a Data-Location Choice

AWS has added cross-Region inference for OpenAI’s GPT-5.6 Sol, Terra, and Luna on Amazon Bedrock in more than 25 AWS Regions. Developers can choose a US geographic inference profile, which keeps processing within predefined US destination Regions, or a global profile that draws on available capacity across supported commercial Regions.

The distinction is operational as much as technical. An application sends requests from a source Region using an inference-profile ID, and Bedrock can process them in a permitted destination Region; global routing may improve throughput under load, but data may cross Regions in the eligible set. AWS says workloads with geographic processing requirements should use the corresponding geographic profile or call a single Region directly.

All three models accept text and image inputs, return text, and offer a one-million-token context window, reasoning mode, server-side tool calling, and prompt caching. Teams can use Bedrock’s OpenAI-compatible Responses or Chat Completions APIs, or its Converse and ConverseStream interfaces, depending on their existing application path.

The deployment choice also reaches access control and data handling. Cross-Region use requires IAM permissions for the inference profile and foundation model in every eligible Region, while CloudTrail records the source Region and an inferenceRegion field for the processing location. AWS also says GPT-5.6 content flagged by automated abuse-detection classifiers may be retained for up to 30 days for offline abuse detection.

For teams already using OpenAI request formats, the integration can be a narrow client change. The harder decision is whether a broader capacity pool fits the workload’s data-location requirements and retention review.