[ENCRYPTED REPORT: SIPHONED TRUTH]

The $38.53 billion headline is the journalistic bait; the operating loss is the structural finding. The non-cash charge of roughly $30 billion attached to the 2025 for-profit conversion (a fair value adjustment of convertible interests and warrant liabilities) is a one-time accounting event that does not reflect day-to-day operations. Strip it out and 2025 net loss is closer to $8 billion — still large, but an order of magnitude smaller than the headline. The operating loss, however, is not a one-time event: it widened from $8.78B in 2024 to $20.92B in 2025, more than doubling year-over-year. On a percentage-of-revenue basis, the operating loss improved only marginally (237% → 160%), and absolute dollars are scaling faster than the relative margin. R&D spend of $19.18B exceeds total revenue of $13.07B — meaning OpenAI is structurally unprofitable on a gross-margin basis before any other cost line. Cost of revenue tripled to $7.5B; sales and marketing quintupled to $5.73B. The story is not that the operating loss exists — early-stage AI labs routinely operate at a loss — but that the absolute dollars are scaling faster than the revenue can chase, even at 3.5x growth. The Microsoft related-party relationship is the S-1 disclosure risk. $10.59B of R&D payments flowed to Microsoft in 2025 (per the audited statements) and the broader $17.2B in cloud and R&D support (per Republic World) makes Microsoft both a major customer and a major supplier. Related-party transactions in an S-1 are a material disclosure: the underwriting banks and the SEC review will treat the $17.2B with particular attention because the same entity is on both sides of the transaction. The S-1 disclosure risk is the load-bearing question for the $852B private-round valuation: if the market prices the IPO at a discount to the private round on the related-party disclosure, OpenAI's last private investors take a mark-to-market loss before the stock even starts trading. The 9-day gap between the confidential S-1 (8 June) and the Zitron/FT leak (17 June) is the secondary story. A confidential filing that surfaces in a journalist's hands within 9 days indicates either a leak by someone with access to the filing (a banker, a lawyer, a competitor, a short seller, an internal whistleblower) or a coordinated disclosure. The leak pre-empts the IPO narrative: by the time OpenAI's bankers can take the company public, the path-to-profitability story has been replaced by the audited-financials story. The 5.6% paid-subscriber-to-WAU ratio (50M / 900M) is the third structural finding: the freemium ceiling is approaching, and the next 10x of growth cannot come from freemium conversion at the current rate. The OSINT verdict: the audited financials show a structurally unprofitable company on a day-to-day basis, scaling R&D faster than revenue, with a $17.2B related-party relationship to the same entity that supplies the cloud it runs on. The $852B private-round valuation will face the audited numbers before it faces the public market.