The infrastructure of AI is being rebuilt faster than the rules that govern it, and the winners are those who can move production workloads into the cloud before regulators catch up. AWS, Cloudflare, and others are redesigning their networks for machine-to-machine traffic rather than human browsing. Anthropic just closed a $65 billion Series H round at a $965 billion valuation, leapfrogging OpenAI and positioning itself for an IPO that will cement the venture-backed AI model as the dominant path to scale. Snowflake is acquiring Natoma to add governance layers for AI agents. Asana is buying StackAI to embed agent builders into workflow software. Visa is investing in Replit to enable agentic payments for developers. The pattern is clear: enterprises are moving past pilots. They're buying tools to manage agents in production, and they're doing it now, before the legal framework solidifies.
Regulation is arriving, but it's fragmented and largely toothless against companies already shipping. Illinois passed a landmark safety testing law with backing from Anthropic and OpenAI, a telling sign that major labs prefer a patchwork of state rules they can navigate over federal enforcement. The EU's AI Regulation is already being violated by all major models according to Aithos's LARA testing tool, which simulates real-world legal exposure. LLMs continue to confidently assert false statements even after explicit corrections, a failure mode that regulatory frameworks assume won't happen. Meanwhile, Elon Musk publicly reframed SpaceX's compute deal with Anthropic as short-term and cancellable, contradicting SpaceX's own S-1 filing describing payments through May 2029. The gap between what companies say about their commitments and what they actually intend to honor is widening precisely as those commitments become the foundation of AI infrastructure investment.
The labor market is already recalibrating around these shifts. Forty-two percent of committed code is now AI-assisted, with roughly 29 percent merged without manual review. H-1B developers are facing a tighter job market as companies redirect hiring toward AI specialists and coding assistants. New mothers returning to software development are entering workplaces transformed by tools they barely recognize. Enterprises are no longer asking whether AI is exciting; they're asking whether it's safe to deploy broadly, which means they're asking for control surfaces, governance, and auditability. IBM and Red Hat are committing $5 billion and 20,000 engineers to Project Lightwell, positioning themselves as the security clearinghouse for open source in the enterprise. The shift from experimentation to production deployment is compressing timelines and concentrating power among companies that can afford to build compliance infrastructure at scale.
Sloane Duvall