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, security, and everyday work surfaces, and by model access that is increasingly routed through managed clouds, compatibility layers, owned product surfaces, and access controls. The infrastructure race underneath is widening into financing, datacenter capacity, packaging, custom silicon, and power. What is particular now is that cloud gatekeeper pressure in Europe is meeting gigawatt-scale buildout and enterprise operating layers at the same time, making control of execution a market-structure question as much as a technical one.

The move

What actors actually did

OpenAI previews GPT-5.6 Sol while Microsoft expands datacenter capacity

The most notable movement this week came from OpenAI, which previewed its new flagship model GPT-5.6 Sol on June 26. The model promises enhanced coding, scientific reasoning, and cybersecurity capabilities, and will be available alongside Terra and Luna variants, each offering different cost and speed options. All versions feature an upgraded safety stack. This marks OpenAI's latest step in advancing high-performance AI across multiple specialized domains.

Microsoft continued its infrastructure expansion, announcing on June 22 the development of a new datacenter campus in Pecos, Texas, expected to add approximately 2 GW to its global datacenter capacity. On the same day, Microsoft and Chevron disclosed plans for a dedicated 2.67-gigawatt natural gas power plant in West Texas, secured under a 20-year power purchase agreement, to supply electricity to Microsoft’s AI and cloud data centers. Microsoft also completed construction of its first datacenter facility in Mount Pleasant, Wisconsin, making it fully operational as of June 23.

Elsewhere, NVIDIA launched the Vera Rubin rack-scale supercomputing platform on June 22, delivering over 7 exaflops of AI performance for scientific workloads, and OpenAI and Broadcom introduced the Jalapeño custom AI accelerator chip, designed for gigawatt-scale LLM inference. Hugging Face announced VLX-Flow, a real-time video understanding model for multimodal interaction, on June 26.

Sources