Marvell bets big on AI connectivity while Oracle raises $20B — buy the AI infrastructure builders

Verdict: AI capex thesis is real, but wait — no entry trigger and insiders are selling

The thesis that AI infrastructure spending is durable has genuine fundamental support — Lumentum's quarter ended June 27, 2026 showed revenue up 24.5% sequentially to $1.01 billion with a 47.4% gross margin, and Marvell's quarter ended August 1, 2026 lifted net margin from 1.4% to 11.2% — but this is a wait for now, on a horizon of roughly the next two earnings cycles. The strongest point for the idea is that the companies receiving the capex are measurably accelerating, not distressed. The strongest point against it is that all three managements were net open-market sellers through the September 30, 2026 provisional filings, and Oracle's free cash flow is negative at −$5.4 billion with more issuance to come. None of the entry conditions is close: LITE trades roughly 19% below its 50-day moving average, MRVL about 29% below, and ORCL roughly 13% below, with ORCL closest on momentum at 2.9 RSI points from its trigger. One scope note: this idea could not be backtested because the Marvell and Oracle raw and adjusted price series use different bases, so no measured win rate or drawdown exists for the entry rules. A Lumentum report showing the $7.16 billion net loss was a one-time charge, or a decisive break below the nearest supports ($750 for LITE, $180 for MRVL, $122.76 for ORCL), would change the assessment either way.

Trade now

The business thesis — that AI infrastructure spending is durable because Marvell is buying Lumentum in an all-stock deal and Oracle raised capital to keep building — is interesting on its own terms, but today's entry conditions are not met, so this is a wait. Every one of the entry rule sets is in waiting mode: each name must first trade back above its 50-day moving average, and then its 14-day RSI, a momentum gauge, must cross above 40 while the MACD line sits above its signal line. None of those reclaim conditions is close to triggering as of the November 2 daily observation. Concrete "wait" means: LITE must reclaim its moving average at $895 (last close $751.73, roughly 19% below), MRVL must reclaim $236.96 (last close $183.16, about 29% below), and ORCL must reclaim $138.41 (last close $122.55, roughly 13% below). The RSI readings are 33.3 for LITE, 23.8 for MRVL, and 37.1 for ORCL — all below the 40 crossover threshold, with LITE and MRVL also below the 45 ceiling the rules allow. The nearest support holds to watch: LITE at $750, MRVL at $180, ORCL at $122.76. A decisive break below those levels would deepen the drawdown, and the idea's own stop framework risks 12% per position with a 60-trading-day time limit on any entry. One scope note: the historical performance of this rule set was not measured because Marvell and Oracle price data used inconsistent raw versus adjusted bases, so no backtest statistics back the entry timing — the trigger levels themselves, not a measured win rate, are what you are following. A pending-risk flag worth weighing against the thesis: insider filings through September 30, 2026 show net open-market selling at all three companies — roughly $96.4 million at LITE, $24.9 million at MRVL, and $45.0 million at ORCL. ### Technical details Entry rule (per ticker, same template): price above the 50-day moving average, 14-day RSI below 45, RSI crossing above 40, and MACD (12,26,9) line above its signal line. Current status per the market state dated November 2, 2026 (daily bars): - LITE: close $751.73 vs the 50-day average at $895.41 — price well below the average. RSI 33.3 (the below-45 filter is met; the above-40 crossover is not, sitting 6.7 points short). MACD is negative and its relationship to its signal line is noted in the live reading. - MRVL: close $183.16 vs the 50-day average at $236.96 — well below. RSI 23.8, 16.2 points below the 40 trigger. - ORCL: close $122.55 vs the 50-day average at $138.41 — well below. RSI 37.1, 2.9 points below the 40 trigger; this is the closest of the three. A second entry template requires the daily low to touch the 61.8% Fibonacci retracement level and then close back above it with an RSI cross above 40; the retracement levels are not supplied, so that condition cannot be evaluated live. Exits: hard stop at −12% unrealized, take profit at +24%, a 60-bar time stop, a close above the second-ranked resistance level, and a close back below the 50-day average. Position sizing is fixed-risk at 2% with a maximum 33% position. Second-ranked resistance targets currently listed: LITE $770, MRVL $170.84 (below the current price, so stale for a long target), ORCL $146.74.

The AI capex flow shows up in these three income statements, not just in headlines

The bull case rests on a real, measurable inflection: all three companies show accelerating revenue, expanding margins or committed capital, which is what the idea's thesis (per the CNBC coverage of the Marvell–Lumentum deal and Oracle's raise) claims the AI buildout is producing. Note first: this idea could not be backtested — historical performance was not measured because MRVL and ORCL raw and adjusted price series use different bases — so no realized return or drawdown statistics exist; the evidence here is fundamental, not trade-tested. ### Marvell's data-center demand is visible in its latest quarter Marvell's quarter ended August 1, 2026 showed revenue of $2.74B, up 13.3% quarter over quarter, with operating margin rising 2.7 points to 16.8% and net income jumping to $308M from $34.5M. That is the causal chain the thesis needs: AI connectivity orders translate into revenue, and volume plus mix lift profitability. Free cash flow held near $479M. If this cadence continues, the Lumentum acquisition adds optical connectivity content to an already-growing franchise rather than rescuing a shrinking one. ### Lumentum has flipped from cash burner to cash generator Lumentum is the strongest single datapoint for the thesis. Its quarter ended June 27, 2026 produced $1.006B of revenue (up 24.5% from $808M the prior quarter), a gross margin of 47.4% — up 3.3 points sequentially after a long climb from the 16–17% trough of fiscal 2024 — an operating margin of 27.8%, and free cash flow of $196M, up 148% sequentially. Among 1,144 Information Technology peers, that operating margin ranks in the 95th percentile. The company that Marvell is buying is accelerating, not distressed, which supports the idea that this is demand-driven consolidation. ### The deal structure and Oracle's raise both point to conviction, not distress Per CNBC, Marvell is paying entirely in stock, meaning no incremental cash drain on a balance sheet funding AI product development, and Oracle filed to sell $20B of stock and bonds specifically to finance AI buildout — hyperscale customers raising capital to keep buying. Oracle's fundamentals corroborate healthy demand underneath: revenue of $19.3B in the quarter ended August 31, 2026, up 0.8% sequentially but with year-over-year growth around 20.6%, a 34.8% operating margin (96th percentile of peers), and operating cash flow of $23.1B. ### What must remain true The bull case requires that Lumentum's margin trajectory continues post-integration, that Marvell sustains double-digit sequential revenue growth…

Scores

• Conviction score breakdown: 49
• Thesis support: 62
• Trade readiness: 22
• Risk quality: 45
• Fundamentals trend: 68

Watch items

• ORCL — Close vs 50-day EMA
• ORCL — RSI (14)
• LITE — Close vs 50-day EMA
• MRVL — Close vs 50-day EMA
• LITE — Support level
• ORCL — Support level