AI expansion is concentrating infrastructure power while losses and labour displacement remain obscured by growth narratives.

Ed Zitron obtained audited OpenAI financial documents, independently verified by the Financial Times. OpenAI posted a $20.92 billion operating loss in 2025 on $13.07 billion in revenue. Total costs hit $34 billion, with $19.18 billion going to R&D alone and $17.2 billion flowing directly to Microsoft. The company has filed a confidential S-1 with the SEC, targeting a public listing Goldman Sachs and Morgan Stanley are leading, at a valuation the financials do not support.

OpenAI is currently valued at $852 billion. Its operating loss ratio improved from $2.37 spent per revenue dollar in 2024 to $1.60 in 2025. R&D spending of $19.18 billion exceeded the entire top line. If you hold index funds through a pension or ISA, an OpenAI IPO at inflated valuation flows into benchmarks. You inherit the risk without being asked.

The Law of the Narcissist: the system insists it is exceptional even as the numbers say otherwise. OpenAI burned $38.5 billion in 2025 and is still pitching a trillion-dollar listing. Retail investors are being primed to believe the growth story is the whole story. The loss column is the story. When you feel urgency around a high-profile IPO, that urgency is manufactured. The product is the hype itself.

Meta is building Meta Compute, an internal cloud unit to sell surplus GPU capacity and AI model access to external companies. Bloomberg broke the story on 1 July 2026. Meta had projected $115 to $135 billion in capital expenditure for 2026. Zuckerberg told employees that 8,000 planned layoffs were a direct consequence of the infrastructure budget. Shares jumped over 8 percent on the news. Stocks of neocloud providers CoreWeave and Nebius were immediately crushed. The excess compute is proof the demand was not there.

Meta raised its 2026 capex forecast to between $125 billion and $145 billion, citing higher component pricing. It simultaneously cut 8,000 jobs to fund that budget. CoreWeave and Nebius, both holding multi-billion-dollar supply contracts with Meta, saw their stocks crater on the Meta Compute announcement. If you work in cloud infrastructure, data centre operations, or AI services, your employer just got a new competitor with near-infinite balance sheet.

The Law of the Landlord: those who own the infrastructure extract rent from those who do not. Meta spent billions building compute it could not fill, then pivoted to charging everyone else for access to it. The layoffs fund the landlord business. The workers who lost jobs are subsidising the cloud margins that replace them. When a company announces a cloud play, ask who paid for the building. The answer is usually the people who no longer work there.

Morgan Stanley research found UK firms with at least one year of AI adoption reported net job losses of 8 percent over the past twelve months, the highest figure among all countries surveyed including Germany, the US, Japan, and Australia, and roughly double the international average. The UK Office for Budget Responsibility modelled the displacement scenario and found AI adoption produces no GDP gain, pushes government borrowing up by £9 billion annually, and cuts tax receipts by £6 billion. The labour market is contracting from the top down.

Youth unemployment in London has reached approximately 25 percent, the highest in the UK. The OBR projects AI displacement cuts UK tax receipts by £6 billion per year and adds £9 billion to annual borrowing costs. That fiscal gap lands on public services. Junior white-collar roles are disappearing first. If you are under 30, entering a profession that relies on analysis, drafting, or coding, the entry-level pipeline is narrowing while senior roles are being protected. The wage floor is moving.

The Law of Entropy: systems degrade quietly before they collapse loudly. The UK is not announcing a jobs crisis. It is absorbing one in silence, distributed across hiring freezes and non-renewals no one calls redundancy. The 8 percent net loss figure is not a headline. It is a structural shift being dressed as transition. When institutions tell you to upskill while removing the rungs you would climb, that is not a plan. That is a managed descent.