Power deals harden while AI rules become operational

The biggest reinforcement is in infrastructure. This week, CleanCore Solutions closed a deal for a 200 MW West Texas AI data center project, Carlyle agreed to sell a $2.6 billion data center power unit to EQT, and MasTec agreed to acquire an electrical contractor focused on data center builds. Put next to Amazon’s reported plan to raise at least $25 billion with AI infrastructure among the uses, the direction is clearer than before: AI capacity is being treated less like a short-lived demand spike and more like something that justifies long-duration financing, power assets, contractor capacity, and site control. The non-obvious part is that these are not only cloud-company moves. Financial buyers, specialist developers, and construction-linked operators are moving in too, while QTS’s canceled Virginia campus shows that local resistance can still slow expansion. That means the contest is spreading across the whole supply system around compute, not only the companies that train models.

A separate movement is regulation becoming more direct and practical. On July 7, the European Commission published an action plan on cybersecurity and artificial intelligence that includes model evaluation capacity and secure access blueprints. On July 9, OpenAI published the GPT-5.6 System Card and launched a Bio Bounty Program, while Anthropic appointed Ben Bernanke to its oversight trust. Read together with the FTC’s proposed policy statement on AI accuracy, the White House push toward model standards, and the Financial Times report on advanced model access via Singapore-based units to blacklisted Chinese-linked subsidiaries, the likely direction is that governance is moving out of general principle and into live mechanisms that affect release practice, documentation, cross-border access, cloud structure, and enterprise conditions of use. That does not tell us how evenly these measures will be enforced, but it does mean companies are already adjusting behavior around them.

The enterprise agent story still matters, but here the picture is more continuity than change. Microsoft expanded Foundry with hosted production agents, connected Foundry IQ into Copilot Studio, and introduced ROI for Agents, while Google launched Cloud Run sandboxes in public preview and set out a route to sell agents through Gemini Enterprise and Google Cloud Marketplace. The near-term meaning remains that vendors are trying to own the managed operating environment around agents - secure runtime, deployment path, governed data access, measurement, and commercial channel - rather than simply offering a chatbot. That still does not mean reliability is solved or that one stack has become the default.