OpenAI files for a mega IPO — ride the wave by buying the AI chipmakers

Verdict: a real AI-demand story, but the entry rules say wait

The strongest argument for this trade is that the demand narrative is showing up in earnings, not just headlines: NVDA's latest quarter (ended July 26, 2026) posted $96.2B of revenue, up 17.9% sequentially with a 66.2% operating margin, and AMD grew revenue 12.5% to $11.5B with net margin jumping to 19.9% from 13.5%. The strongest argument against is the ownership record: filings for the quarter ended June 30, 2026 show net open-market insider selling at all three names — about -$565.4M at NVDA, -$153.2M at AMD and -$19.5M at MRVL — roughly $738M combined, plus a live paper track sitting at $9,906 against a $10,000 start (-0.9% over 10 trades). The verdict would flip if AMD's RSI cools back into the 50–70 band (it closed at 72.7, only 2.7 points over the ceiling) while ADX stays above 20, or NVDA's ADX (11.8) rises above the 20 threshold — the setup's own entry conditions. Fundamentals remain healthy but cash conversion is softening at the margin (NVDA free cash flow down 56.0% sequentially to $21.4B; AMD's down 39.3%), and no robust parameter setup was established by the sensitivity evaluation. Wait until the rule set confirms; nothing is actionable today.

Trade now: two near-misses, no entry yet

## Trade now — waiting, with two distinct blocks to watch Nothing is actionable in NVDA or AMD today, and "wait" has a precise meaning here. AMD closed at $520.22 with RSI (14) at 72.7 — above the 70 ceiling, so the momentum-band entry (RSI between 50 and 70) is violated by just 2.7 points; ADX (14) at 31.9 already clears the 20 hurdle. The pullback-with-support variant additionally needs the daily low to touch the nearest support level at $520 while closing above it, which is genuinely close to in range. NVDA closed at $224.17 with RSI (14) at 52.0, comfortably inside the 50–70 band, but ADX (14) at 11.8 is 8.2 points short of the 20 threshold — the stock is consolidating, not trending, and the rule set correctly stands aside. MRVL is not a tradable leg, but for context it sits at $207.95 with RSI at 37.8 and is already below its 10-day average at $224.95, so it would trigger the exit condition rather than an entry. Risk framing is simple: the strategy uses a fixed 2.5% stop loss per position against a 5.0% take profit, an effective reward-to-risk of 2:1, with positions capped at 25% each. The realized paper track since July 14 shows 10 trades and a -0.9% total return with a 10.7% maximum drawdown; the realized backtest over 60 months returned 79.3% across 198 trades with a 44.4% win rate and a 28.6% maximum drawdown — consistent with a 2:1 payoff at a sub-50% hit rate. Respect the geometry: if AMD's RSI cools into the band while price holds above $520, that is the cleaner of the two setups. If NVDA's ADX climbs above 20 while RSI stays in-band, the trend leg activates. Until one of those flips, do nothing. ## Catalysts ## What flips each entry — a timeline The nearest potential trigger is AMD: RSI (14) only needs to shed 2.7 points from 72.7 to print at or below 70 while ADX (14) holds above 20 — plausibly within days of any consolidation day. NVDA is further out: ADX (14) at 11.8 must nearly double to reach 20, which typically follows a directional breakout rather than the current range. The exit side matters just as much once a position is live: NVDA closed only 0.7% above its 10-day average at $223.45, so a single weak close there would fire the exit almost immediately after entry — a real whipsaw risk. Fundamentally, the data supports the bullish thesis but flags caution. NVDA's latest quarter (period ended July 26) showed revenue of $96.2B, up 17.9% sequentially, with a 62.0% net margin; AMD grew revenue 12.5% to $11.5B with net income up 66.1%. But ownership filings for the period ended June 30 show net open-market insider selling at all three names — about -$565.4M at NVDA and -$153.2M at AMD — a bearish overhang worth watching. The next SEC quarters for AMD (fiscal quarter after June 27) and NVDA (after July 26) are the scheduled catalysts where revenue and margin trends get re-confirmed or broken.

The 'picks and shovels' thesis has numbers behind it, not just IPO hype

The idea argues that OpenAI's confidential IPO filing — confirmed by CNBC on June 8, 2026 — should pull money into the chipmakers that make AI physically possible, and the fundamentals at the three targeted names support that demand narrative. NVDA's most recent quarter (ended July 26, 2026) shows revenue of $96.2B, up 17.9% sequentially, with a 66.2% operating margin and $59.7B of net income. AMD's quarter ended June 27, 2026 shows revenue of $11.5B, up 12.5% from the prior quarter, with net margin expanding to 19.9% from 13.5%. MRVL's quarter ended August 1, 2026 shows revenue of $2.7B, up 13.3%, with net income jumping to $308M from $34.5M and operating margin rising to 16.8% from 14.0%. These are accelerating earnings, not stale momentum. The rule-based setup agrees with the thesis directionally: enter long when the sector (SOXX) closes above its 10-day moving average with RSI(14) between 50 and 70, and exit on a close below the 20-day average or a 5% trailing stop. That is a momentum-with-a-seatbelt design — it only buys strength while the broader SOXX uptrend is intact, which is exactly the environment an AI IPO narrative creates. Per the Yahoo Finance piece from June 9, chipmakers were expected to power a further rebound on the news, and these names are not niche plays: AMD (8.1%), NVDA (6.8%), and MRVL (4.9%) together are roughly a fifth of SOXX's weight. The realized backtest evidence is genuinely strong…

Scores

• Conviction score breakdown: 60
• Thesis support: 72
• Trade readiness: 45
• Risk quality: 55
• Backtest evidence: 60
• Fundamentals trend: 70

Watch items

• AMD — RSI (14)
• AMD — Support touch
• NVDA — ADX (14)
• NVDA — Close vs 10-day SMA
• NVDA — Insider net open-market value (June 30 filings)
• AMD — Quarterly revenue