The AI Crisis

— OpenAI and Anthropic are deeply unprofitable despite massive customer bases, with OpenAI losing $12 billion in Q1 2026 on $5.7 billion revenue and facing $800 billion+ in obligations by 2030.

— Both companies are heavily subsidized on compute costs and still burning capital; their accounting shows they lose more money with each additional customer.

— China's open-source AI models (e.g., Kimi K3) are closing the capability gap to US frontier models in 2–4 weeks while costing a tenth as much, eroding any competitive moat.

— AI industry leaders are engineering regulatory panic around "dangerous AI" and lobbying for government partnerships, liability protection, and regulatory boards to preserve their business model and block competition.

— The goal is a centralized surveillance infrastructure with government backing, guaranteed customers, impossible-to-enter compliance barriers, and immunity from liability—funded by taxpayers and users.

— Mark Zuckerberg's counter-argument: alignment and liability are already handled by free-market customer choice and corporate responsibility without heavy-handed regulation.

— Recognize that regulatory "safety" campaigns from AI labs are primarily defensive moves by unprofitable companies seeking subsidies, monopoly protection, and government partnerships rather than genuine public interest advocacy.

— Advocate for open-source AI development and resist regulatory capture that would concentrate AI power in a handful of corporations and governments, since competitive alternatives already exist and improve rapidly.