The companies building the most consequential technology in history are simultaneously pricing it, policing it, and pocketing the proceeds.

Anthropic filed a confidential S-1 with the SEC in June 2026 and is targeting an October listing at $2 trillion, more than double its May private valuation of $965 billion. Bankers are underwriting revenue projections of $100 to $120 billion by December — figures that are investor expectations, not company guidance. The $2 trillion ask prices in growth that has not happened yet, at a 33x revenue multiple, on a business with a 5.1 percent operating margin.

At $2 trillion, Anthropic would be the largest IPO in history, surpassing SpaceX's $85.7 billion raise. The Wall Street Journal reported the offering could seek up to $100 billion, more than the entire US IPO market raised in any year outside 2020 and 2021. UK and US retail investors via index trackers and pension funds will absorb this stock at whatever price institutional allocators set. The valuation rests on a December revenue run rate that does not yet exist.

Law of the Narcissist: the belief that one's own story is so exceptional that the normal rules of evidence do not apply. When a company quintuples its valuation in six months with no annual profit on record, the pitch is not a business case. It is a mirror. The public is being asked to see itself in the reflection — to fund a future that the insiders are already cashing out of. An 18-year-old buying this IPO on day one is not investing. They are the exit.

OpenAI, Anthropic and Google have been meeting since July 2026 on a self-regulatory AI standards body modelled after FINRA — the US financial industry watchdog. The body would test and audit frontier models before public release. The plan requires a narrow antitrust waiver allowing direct competitors to coordinate. Cohere CEO Aidan Gomez publicly named it a cartel. Former OpenAI executive Zack Kass called it regulatory capture designed to protect incumbents from open-source competition. The White House met all four labs in August to discuss a parallel federal cybersecurity framework.

The mechanism is precise: mandatory compliance audits are fixed costs. A large lab absorbs them. A startup or open-source project cannot. Audit-based regulation at the frontier level quietly reclassifies mid-market and challenger tools as non-compliant over time. For UK and US workers whose employers rely on cheaper open-source AI infrastructure, the downstream effect is higher software costs, fewer competitive tools, and a handful of US labs with pricing power over every business that cannot build its own model.

Law of Projection: the act of loudly naming a danger you are creating. Anthropic has donated $40 million to political groups advocating government AI safeguards while preparing a $2 trillion IPO — a regulatory moat is the cheapest competitive advantage available to a company that has never turned an annual profit. When the people building the risk propose themselves as the people managing the risk, the safety language is doing commercial work. An 18-year-old must ask: who benefits when the rules are written by the companies the rules are meant to govern?

Jacob Coxon resigned from Anthropic this month, warning that AI labs are gambling with human lives. Former Google DeepMind researcher Bilal Chughtai separately stated he believes AI has the potential to kill everyone. Anthropic's response was to name METR — Model Evaluation and Threat Research — as its proposed independent safety evaluator. METR acknowledged that during a prior evaluation, at least six participants had close personal relationships with staff at the companies being evaluated, and that no formal conflict-of-interest policy existed at the time.

METR raised $71 million in August 2026 from philanthropic foundations linked to Anthropic's investor networks. METR's founding is connected to Paul Christiano, who worked with Dario Amodei at OpenAI and served as one of the first trustees of Anthropic's Long-Term Benefit Trust. For UK and US households, this is the entire safety architecture being offered in place of government regulation: a nonprofit funded by adjacent money, staffed by adjacent people, checking the work of companies preparing to raise $100 billion from the public.

Law of the Trap: the exit route is controlled by the same party that set the trap. You are told the technology is dangerous. You are told only the people building it understand it well enough to regulate it. You are told to trust their chosen auditor. The auditor shares funders, alumni and access with the companies it audits. Every step leads back to the same room. An 18-year-old must understand: when the alarm and the fire extinguisher are sold by the same person, the incentive is not to put the fire out quickly.