What stands out to me is not just that Micron’s numbers are explosive, but that the shape of the explosion is fundamentally incompatible with the current valuation unless one assumes a near-permanent rewrite of memory economics. In seven reported quarters, revenue rose from $7.75B to $41.46B, while net income went from $887M to $28.24B. That is not normal operating leverage; it is an extreme mix of pricing, utilization, and likely high-value product concentration. The annual series tells the real story: FY2023 was a -$5.83B net loss on $15.54B of revenue, FY2024 recovered to just $778M of profit on $25.11B, and FY2025 jumped to $8.54B on $37.38B. This is still a deeply cyclical manufacturer with violent earnings elasticity, not a software-like compounding machine. The balance sheet is healthy enough — $9.64B cash against $12.09B debt and only 0.22x debt/equity — so this is not a solvency short. The problem is that the market cap of $834.62B is asking investors to capitalize a cyclical surge as though Micron has escaped the cycle.

The cash flow profile is the biggest tell. FY2025 operating cash flow of $17.53B looks strong until you put it next to $15.86B of capex, leaving only $1.67B of free cash flow. On an $834.62B market cap, that is essentially no cash yield at all. This matters more than the spectacular income statement because memory has to keep spending to stay relevant; capex is not optional growth vanity spend, it is the cost of remaining in the game. The latest quarterly annualized run-rate is obviously much stronger than FY2025, so trailing FCF understates current power, but even then the business has to prove that these margins can survive once competitors react. A 68.1% quarterly net margin on $41.46B of revenue is so far outside Micron’s own history that I treat it as a peak-condition datapoint, not a base-rate anchor. Even if one grants a structurally improved product mix from HBM and AI-driven DRAM, paying 22.2x sales and 45.6x EV/EBITDA for a memory producer is an aggressive statement that this is a new species of semiconductor company. I do not believe that.

I agree strongly with Opus’s core claim that the stock is overvalued because peak-cycle economics are being treated as durable. When Opus says the company is “priced as if Micron were TSMC with Nvidia’s growth and Apple’s margin stability,” that is directionally correct. I also agree with the attack on the classifier: calling Micron a “narrative_platform” is obviously nonsense and leads to bad metric handling. This is exactly the kind of company where trailing profitability metrics need contextualization, not dismissal. I also agree that the annual history itself refutes any easy permanence thesis: a business that went from $8.69B net income in FY2022 to -$5.83B in FY2023 and back to $8.54B in FY2025 is telling you, loudly, that normalized earnings are far below whatever today’s run-rate implies. Opus is also right that the insider sales are not bullish, though I would not make much of ten tiny same-day sales without dollar values; they are a faint yellow flag, not evidence.

Where I disagree with Opus is mostly in the confidence of the downside math, not the direction. First, Opus leans hard on “a 68% net margin on $41B of quarterly memory revenue is not something the memory industry has ever produced,” which is fair as a warning, but the better argument is not historical incredulity alone; it is that the capital intensity remains brutal even in the upturn. I would put more weight on the disconnect between earnings optics and owner earnings than on the idea that the latest quarter might be a data artifact. Second, I think Opus’s fair-value framing of roughly $200-320 per share is too punitive given the possibility that HBM and AI memory have genuinely raised Micron’s mid-cycle floor. The FY2025 income statement already shows the business can earn $8.54B before the latest supernormal quarters fully annualize, and if a structurally better mix allows normalized earnings power closer to, say, $20-25B rather than $15-20B, then a fair cyclical valuation can be above Opus’s range without remotely justifying $739. I also disagree with the implication that this is a clean short just because the setup is textbook. With a balance sheet this solid and momentum this extreme — recent revenue growth of 48.9% YoY and earnings growth near 998% — timing a short in a supply-constrained semiconductor upcycle can be lethal even when the thesis is right.

A careful skeptic of both my view and Opus’s would say we are overfitting Micron’s past to a future in which memory has become more oligopolistic and more strategic to AI systems than at any prior point. They would note that annual figures lag the present, that FY2025 FCF captures capex before the full benefit of the latest revenue surge, and that if even a portion of the $41.46B quarterly revenue base is tied to longer-duration, high-margin HBM demand, then “through-cycle” may genuinely be higher than the historical analogs suggest. That skeptic has a real argument. But they still have to bridge from “higher through-cycle earnings” to an $834.62B market cap, and the supplied numbers do not get them there.