The numbers tell a story the models are only partially capturing. Revenue went from $2.1M in 2021 to $130M in 2025 to roughly $247M on a trailing-twelve-month basis through June 2026 — a 118x expansion in four years, with the most recent quarter at $80.1M, up 370% year-over-year from $20.7M a year earlier. That growth is real, booked, and accelerating in absolute dollars. But the single most important data point in this file is the one the thesis evaluation and valuation synthesis underweight: gross margin has collapsed from 74% in 2022 to 63% in 2023 to 52% in 2024 to 40% in 2025 to 29% on a TTM basis. That is not a cyclical dip; that is a structural deterioration in the unit economics of selling quantum hardware at scale. A company whose revenue is growing 370% but whose gross margin is falling 500 basis points a year is not building a moat — it is buying revenue with margin. The TTM operating margin of -408% and net margin of -553% are not "pre-profit" noise; they are the arithmetic consequence of a cost structure that does not scale with the top line. The quarterly net income swings of +$805M in Q1 2026 and -$1.87B in Q2 2026 on $65-80M of revenue are clearly non-operating items — warrant revaluation, stock-based comp, or mark-to-market on equity positions — and should be stripped from any fundamental read. The annual 2025 net loss of $510M on $130M revenue is the honest number, and it implies a burn rate that the $1.03B cash pile covers for roughly three and a half years at the 2025 operating cash flow of -$283M, less if burn accelerates with headcount and fab commitments.

The valuation math is where the models split from me. The pre-flight and thesis evaluation anchor on 123x P/S using the 2025 annual revenue of $130M, which is stale. On TTM revenue of ~$247M, the multiple is closer to 65x. Still, 65x revenue for a company with 29% gross margins and no EBITDA is not a "reasonable" multiple by any framework I apply. The FCF quality flag of "Poor" is correct and understated: free cash flow of -$300M against a $16B market cap means the market is pricing in a future where this company generates meaningful cash, and the timeline for that is genuinely uncertain. The "Revenue Confidence: Low (decelerating)" flag is partially wrong — Q2 2026 revenue of $80.1M was up 24% sequentially after Q1's anemic 4.5% gain, so the deceleration narrative is one quarter old and already reversing. The insider "Net Buying" signal is a vesting-event artifact: the 22,630 shares in "A-Award" transactions on June 17 are almost certainly RSU vestings, not new grants, and the 9,329 shares sold on June 18 at $39.52 (a 53% drawdown from the $84.64 high) are modest in size but directionally consistent with insiders taking liquidity after a correction. I would not read either as a strong signal in either direction.

The contrarian case, even with every bull data point in front of me, is this: you are paying $16 billion for a company whose entire value proposition rests on trapped-ion quantum computing achieving fault-tolerant commercial utility within a 5-to-10-year window, that IonQ's architecture beats superconducting (IBM, Google) and photonic approaches, and that IonQ specifically — not a hyperscaler with $100B+ in annual R&D budgets — is the entity that crosses the threshold. The gross margin trajectory is the tell. If the business model required 74% gross margins to be viable and is now at 29%, the path to profitability is not a matter of revenue scale but of a fundamental shift in product mix toward software, networking, or services — none of which are yet material to the P&L. The "NVIDIA moment" narrative is seductive but historically incoherent: NVIDIA had 20 years of GPU architecture iteration, a mature semiconductor supply chain, and a clear product-market fit in gaming before the AI inflection. IonQ has none of that scaffolding. The $1.03B cash and zero debt are genuine strengths — they eliminate near-term dilution risk and give the company a real runway — but they do not change the fact that the stock is 100% narrative, as the market-forces layer correctly identifies.

The synthesis verdict of "High Conviction Required" is a non-verdict dressed as one. It acknowledges the revenue growth is real, the physics is real, and the multiple is extreme, then refuses to commit. I can do better. At $39.52, the 53% drawdown has already removed a substantial portion of the speculative froth that was present at $84.64. The stock is not a screaming buy — there is no cash flow to anchor a DCF, and the margin collapse is a genuine structural risk that the bull case hand-waves. But it is not a screaming sell either, because the revenue trajectory is the fastest I have seen in a hardware company of this size, and the balance sheet is clean. I rate it fairly valued to modestly overvalued at the current price, with the key variable being whether gross margin stabilizes above 35% in the next two quarters. If it continues the 500-basis-point-per-year decline, the 65x multiple is indefensible and the stock should trade in the $20s. If it stabilizes and the revenue growth holds above 200% YoY, the current price is a reasonable entry for a high-conviction quantum allocation. The models are directionally correct on the risk but wrong to treat the P/S multiple as the primary risk metric when the margin trajectory is the leading indicator.