AI CAPEX IS HITTING $1 TRILLION. NOW CORPORATIONS WANT TAXPAYERS TO UNDERWRITE THE INFRASTRUCTURE. When private monopoly runners ask global governments to regulate them, look at the balance sheets. Global AI capital expenditure is projected to surpass $380B this year. Free cash flow conversion across major hyperscalers has dropped from 65% to under 38% as GPU depreciation cycles compress to 36 months. Private venture capital and corporate cash reserves cannot sustain the projected $1.4T grid and compute buildout alone. Appearing before the UN Security Council under the banner of "existential risk" serves two specific macroeconomic objectives: 1. Regulatory Moat. Enforcing mandatory licensing and non-proliferation thresholds on training runs above 10^26 FLOPs legally eliminates open-source alternatives. It criminalizes low-cost competitors before they can compress pricing power. 2. Sovereign Underwriting. Constructing 5GW to 10GW dedicated nuclear and grid facilities requires sovereign debt guarantees. By framing frontier models as national security assets equivalent to uranium enrichment, corporations shift the capital expenditure burden to public balance sheets. This is not safety diplomacy. This is classic infrastructure capture designed to socialize trillions in development costs while privatizing terminal margins.