Not all AI trades are the same: memory makers like Micron are enjoying a genuine, physical price surge — memory costs up over 500% — which shows up as real revenue now, not promised revenue later. At the same time, the highest-profile AI names are under a

Verdict: Micron's pricing power is real, but wait for the entry signal

The strongest point for this trade is that Micron's pricing-power thesis is showing up in reported numbers: fiscal 2025 revenue of $37.4B, up 48.9%; gross margin climbing from 39.8% to 74.4% by the February 2026 quarter; free cash flow of $17.6B in the latest quarter; and a dividend lifted to $0.53 trailing per share with 15.2% annual growth — real cash today, not promises. The strongest point against is that the filings are unambiguously adverse: as of the June 30, 2026 reporting period, insiders were net open-market sellers across all three names, including roughly $231 million at Micron itself, so the smart money does not appear to be rotating in. The completed nine-month backtest returned 3.5% on four trades with a 7.7% maximum drawdown — positive but thin, and no robust parameter setup was established, so the published settings stand as-is. Meanwhile, none of the three tickers has met its full entry conditions: Micron at $977.30 still sits about $19.64 above its $957.66 Bollinger middle band, with the 50-day EMA condition even further away. The verdict is wait — the thesis is credible, but the entry discipline and the September 30 insider-filing update should decide timing. A further hot memory-price disclosure, or a filing showing insider selling fading, would materially strengthen the case.

Trade now: The rotation trade is live on the thesis, but the entry rules say wait

The idea argues Micron has real pricing power today — memory costs up over 500% — while Nvidia and Palantir face a double squeeze from a $1.1 billion disclosed bet against them and borrowing costs at a 19-year high (per the idea's own thesis and the Michael Burry disclosure it cites). The numbers support the rotation narrative: Micron's annualized return over the 730-day lookback is about 122% versus roughly 33% for Nvidia, and the completed nine-month backtest on this setup traded four times, won half, and returned about 3.5% with a maximum drawdown of 7.7%. But none of the three tickers is an entry today, and that is the setup working as designed, not a trust problem. On Micron at $977.30, two of four entry conditions are already met — RSI (14) is 55.5, above its 45 floor, and ADX (14) is 25.8, above 20 — but price needs to fall about $19.64 to at or below the Bollinger middle band at $957.66, and the 50-day EMA at $933.29 needs to sit above price, meaning the stock is roughly $44 too high relative to that condition. Palantir has actually cleared its Bollinger condition (price $175.59 versus the band at $175.68), but its ADX of 15.1 is well short of 20 and the EMA condition is not close. Nvidia at $219.40 is nearly on its band (about $0.43 away) but its ADX of 7.3 is far from the 20 threshold. So "wait" means concretely: do nothing until all four conditions align on a single name. The risk math makes patience cheap. Once an entry triggers, the strategy sizes the position at fixed 2.6% risk per trade (positions capped at 25% of the book) and exits at a 5.1% gain — roughly 2-to-1 effective reward-to-risk — with additional signal exits if RSI pushes above 75 with a negative MACD histogram, or after 60 bars. Note one factual scope limitation: robust parameter optimization could not be run because the historical data coverage was incomplete, so the published settings are the frozen original configuration rather than a tuned variant. If you want the rotation exposure today, understand you are trading ahead of your own signal. The disciplined play is to set alerts at the Bollinger middle bands — $957.66 on Micron, $218.97 on Nvidia, $175.68 on Palantir — and let the EMA and ADX confirmations decide which name triggers first.

Micron has the earnings that the AI story stocks only promise

The idea's core claim — that Micron is converting an AI-driven memory price surge into revenue today rather than promised revenue later — is directly visible in the numbers. Micron's fiscal 2025 (ended August 28, 2025) delivered $37.4B in revenue, up 48.9% year over year, with $8.5B in net income and $17.5B of operating cash flow. The pricing-power narrative strengthens through the fiscal year: quarterly gross margin climbed from 39.8% for the full year to 56.0% in the November 2025 quarter and 74.4% in the February 2026 quarter, exactly the shape you'd expect if memory pricing were genuinely repricing upward rather than being capitalized into a distant forecast. The catalyst cited in the thesis is corroborated by the news tape: per the Yahoo Finance piece from September 17, 2026, Micron jumped 5% after Intel's CEO said memory prices have surged more than 500%. That is a same-day, physical-cost-driven repricing — the mechanism the thesis says separates MU from the story stocks. Meanwhile, the same day's coverage of Michael Burry's roughly $1.1 billion put position against Nvidia and Palantir, and the September 15 report that the 10-year yield hit a 19-year high, supply the pressure side of the rotation: rate-sensitive, future-promise valuations face a double squeeze. The backtest evidence, while modest in sample, does not contradict the thesis. Over a nine-month daily window the MU-led rules-based strategy completed 4 trades with a 50% win rate, returning 3.5% with a maximum drawdown of 7.7%. That is a positive, controlled result rather than a moonshot — consistent with a rotation trade that grinds rather than explodes. Cash quality also favors the bull case on a relative basis. Micron holds $9.6B in cash against $11.5B of long-term debt, a current ratio of 2.5, and just returned to positive free cash flow ($1.7B in FY2025 after years of capex-heavy losses) with free cash flow running at $3.0B, $5.5B, and then $17.6B in the three most recent reported periods. Management is confident enough to have lifted the dividend to a $0.53 trailing per-share rate after years at $0.115 quarterly — a 15.2% annual…

Scores

• Conviction score breakdown: 60
• Thesis support: 78
• Trade readiness: 55
• Risk quality: 48
• Backtest evidence: 45
• Fundamentals trend: 75

Watch items

• MU — Bollinger (20) middle band
• MU — EMA (50) vs price
• MU — ADX (14)
• MU — RSI (14)
• NVDA — ADX (14)
• NVDA — Ownership filing (Q3 2026)
• PLTR — Bollinger (20) middle band
• PLTR — ADX (14)
• PLTR — Next quarterly earnings
• MU — RSI (14) exit level