streugut.ch
Archive

The direction

Where the AI race is heading

AI competition is moving away from frontier models alone toward control of the stack that turns models into governed execution. That shift is being carried by agent environments tied to enterprise identity, data, workflow, and security, and by model access that is increasingly routed through managed clouds, compatibility layers, owned product surfaces, and now access controls. The infrastructure race underneath is widening into financing, datacenter capacity, and energy planning. What is particular now is that enterprise rollout is being organized as a governed operating layer just as who may use advanced models is tightening along regulatory and geopolitical lines.

The move

What actors actually did

Microsoft’s Copilot Cowork launches, OpenAI expands model testing, xAI integrates Grok with Amazon Bedrock

Microsoft moved its Copilot Cowork AI agent, developed with Anthropic, from preview to general availability for Microsoft 365 Copilot customers, now with usage-based pricing. This transition means more organizations can directly deploy collaborative AI within enterprise environments, marking a clear step for Microsoft in broadening its AI agent offerings.

OpenAI launched Deployment Simulation for GPT-5 series models, giving developers a new way to test these models by simulating realistic conversation contexts and identifying potential unwanted behaviors before full-scale rollout. Meanwhile, xAI made its Grok models available on Amazon Bedrock, letting AWS customers access Grok’s advanced reasoning, coding, voice, and multimodal capabilities through managed AI infrastructure.

Anthropic opened its Seoul office, partnering with Korean enterprises like NAVER and Nexon to deploy Claude Code, and rolling out solutions at LG CNS, Hanwha Solutions, and Samsung SDS. Microsoft also continued its global push, with its AI business in China growing and ByteDance expected to spend over $1 billion annually on Microsoft AI and cloud services.

Sources