The legal and regulatory machinery designed to govern AI is grinding slower than the technology it's meant to constrain, and that gap is creating space for market consolidation disguised as business-as-usual. Anthropic's $1.5 billion copyright settlement is stalled because lawyers are fighting over $320 million in fees while the underlying question of AI training liability remains unresolved. The Musk v. Altman trial just concluded without settling whether OpenAI breached its nonprofit mission, leaving the structure of AI governance to corporate reorganizations rather than courts. Meanwhile, arXiv is banning AI-generated papers with yearlong suspensions, but this is moderation by exclusion, not resolution. The real action is happening in places where rules haven't caught up: Apple integrating ChatGPT without OpenAI's satisfaction, AWS layering prompt optimization onto Bedrock, Runway betting on video generation as the path to world models. The system isn't broken. It's working exactly as designed for the companies that can afford to operate across multiple jurisdictions and wait out litigation.
Energy is becoming the new constraint, and it's reshaping where compute happens and who controls it. Pennsylvania towns are holding meetings against data center booms. Lake Tahoe faces higher electricity prices as AI workloads spike demand. Reno, Nevada just became the first local government in the state to pause new data center applications. These aren't abstract policy debates. They're infrastructure fights with real costs, and they're happening faster than capacity markets can absorb them. Anthropic's arrangement with SpaceX to secure compute capacity outside the traditional hyperscaler duopoly signals that the smart money sees this constraint as permanent, not temporary. AWS, Microsoft, and Google have spent years training users to assume infinite elastic capacity. That assumption is breaking. Companies are now negotiating directly for power and silicon the way they used to negotiate for office leases. The hyperscalers still own the data centers, but they're no longer the only path to scale.
The boundary between product and regulation is collapsing, and companies are choosing which side they want to be on. Microsoft faces a UK antitrust probe into AI lock-in and bundling. Illinois is drafting an eight-bill AI regulation package modeled on California and New York frameworks. Google is hiring 1,513 forward-deployed engineers to help enterprises scale AI deployments, which is both market expansion and regulatory preparation. OpenAI launched a personal finance product that connects to bank accounts, which is either fintech innovation or financial services regulation waiting to happen. Anthropic's Claude Code product lead says they have no grand plan, by design, which is a statement about avoiding the appearance of monopoly more than about actual strategy. The companies that move first into new verticals and geographies while regulation is still drafting win the right to define what compliance looks like. The companies that wait for clarity lose the market. This is why you're seeing so much product velocity right now, three days before Google I/O: the window between adoption and regulation is closing, and the companies shipping today are the ones that will write the rules tomorrow.
Sloane Duvall