AI boom shifts from Big Tech to silicon — momentum play on secondary chipmakers and power suppliers

Verdict: the AI hardware rotation is real, but the trade isn't live yet

The idea's core thesis — money rotating from Big Tech platforms into AI hardware suppliers — is well supported by the fundamentals: AMD's gross margin rose from 52.8% to 53.8% quarter over quarter with net margin jumping from 13.5% to 19.9%, and Micron's fiscal Q3 2026 printed an 84.6% gross margin on $41.5B of revenue with $17.6B in free cash flow. The strongest point for the trade is that the completed nine-month backtest delivered a 41.4% return across four trades with a 75% win rate and a contained 15.4% maximum drawdown. The strongest point against is that ownership filings through June 30, 2026 show net open-market insider selling across all three names — roughly -$153.2M in AMD, -$25.0M in Bloom, and -$231.1M in Micron — a standing bearish backdrop at what CNBC reported as record levels. Critically, no entry condition is confirmed today: AMD's 20-day average is still $15.56 below its 50-day, and Bloom and Micron's crossovers are flagged near rather than confirmed. The longer evaluation windows could not run because Bloom Energy daily data coverage was incomplete, so there is no evidence the setup works beyond nine months and no robust parameter setup was established. Verdict: wait — set alerts at the crossover levels and reassess if AMD's moving averages confirm the cross.

Trade now: set alerts on the 20/50-day crossovers — nothing is entry-ready yet

The strategy enters long MU, AMD, and BE on the daily chart when price closes above the 20-day average and the 20-day average crosses above the 50-day average. As of the latest close, none of the three names has both conditions live. AMD closed at $516.13, above its 20-day average of $481.94, but that 20-day average sits about $15.56 below its 50-day average of $497.50 — the crossover has not happened. BE closed at $275.75 with its 20-day average ($227.97) already above its 50-day average ($225.79) by $2.19, and MU closed at $975.39 with its 20-day average ($964.65) above its 50-day average ($928.73) by $35.92; both are in or near the crossover zone, but the cross is flagged as near rather than confirmed. What waiting means concretely: no positions today — set alerts at the crossover gap levels and check each daily close before acting. Risk is defined by the strategy itself: a stop at a 2.4% loss per position and a take-profit at a 4.8% gain, roughly a 2:1 reward-to-risk. Sizing is fixed-risk with a 25% maximum per name. One caveat from the evidence: stops and targets were filled on daily bars rather than intrabar, so realized exit quality in the test record is approximate. The completed backtest evidence supports this immediate framing: over the most recent nine-month window, the MU leg traded four times, won 75% of trades, returned 41.4%, with a maximum drawdown of 15.4%. The longer 12-, 24-, and 60-month windows could not be completed for the same reason the optimization run stopped — BE daily data coverage was incomplete — so no robust parameter setup was established, and the frozen rule set is what you would trade as written.

The 'picks and shovels' rotation has the numbers behind it

The idea argues that AI money is rotating from Big Tech platforms into the hardware suppliers, and the most recent fundamentals bear that out. Micron's fiscal Q3 2026 (ended May 28, 2026) shows revenue of $41.5B with a gross margin of 84.6% and net margin of 68.1% — an extraordinary step up from the 74.4% gross margin in the prior quarter and the 39.8% gross margin posted for the fiscal year ended August 2025. Free cash flow in the same quarter hit $17.6B. That is exactly the profile you'd expect from a memory supplier with pricing power at the center of an AI build-out, and it validates the strategy's trigger stock (MU's 20/50-day moving-average cross drives entries). AMD supplies the second leg of the rotation, and its momentum is equally visible in the trend data. The most recent two quarters show accelerating quality: gross margin rose from 52.8% in the quarter ended March 28, 2026 to 53.8% in the quarter ended June 27, 2026, net margin jumped from 13.5% to 19.9%, and operating margin improved from 14.4% to 17.3%. Quarterly revenue reached $11.5B in Q2 2026, up from $10.3B the quarter before. For the fiscal year ended December 2025, AMD grew revenue 34.3% year over year and generated $6.7B in free cash flow — placing it in the top 1% of its sector on that metric. The power-infrastructure leg is where the cited news does the heavy lifting. Reuters reported on June 30, 2026 that Bloom Energy and Brookfield expanded their AI infrastructure power partnership to $25 billion — a concrete, contracted demand signal for the thesis that data-center electricity needs are accelerating. Bloom's own fundamentals support a turnaround narrative too: it posted positive net income of $73.7M in the quarter ended March 31, 2026, its most recent observation, after a full-year net loss of $87.1M for 2025, and full-year 2025 revenue grew 38.9% year over year. On the evidence read itself: this is a completed backtest, not a theoretical setup. Over a nine-month daily evaluation, the MU-led pair produced a 41.4% return across 4 trades with a 75% win rate, while the worst peak-to-trough equity drawdown was 15.4%. A majority of trades profitable with a double-digit return and a drawdown contained to roughly half the gross gain is a historically coherent risk/reward for a momentum structure — and it lines up with the CNBC reporting from June 30, 2026 that Micron, Intel and…

Scores

• Conviction score breakdown: 63
• Thesis support: 78
• Trade readiness: 52
• Risk quality: 55
• Backtest evidence: 55
• Fundamentals trend: 76

Watch items

• AMD — SMA (20) vs SMA (50) crossover
• BE — SMA (20) above SMA (50) by
• MU — SMA (20) above SMA (50) by
• AMD — ROC (14)
• MU — Net insider open-market value
• AMD — Net insider open-market value
• MU — Next dividend ex-date
• AMD — Gross margin, Q2 2026