Skip to content

Dossier · MRVL · Dormant

MRVL · Marvell Technology, Inc. · Stock research

LOW Cyclical recovery Catalyst · ai-chips-memory

Last analysed ·

Current thesis

Custom-silicon ASIC royalty story intact, but the AI-semi theme is mid-correction chips posted their worst month vs software on record (7/17) with money rotating into energy. Buying a falling sector with no confirmed higher low is the trap; wait for the the published invalidation level-255 May shelf to hold before re-engaging.

Invalidation trigger

a daily close that breaches the risk threshold loses the May breakout shelf and rising weekly 20-EMA. Secondary breaks: Broadcom disclosing incremental custom-ASIC share at AWS Trainium or Microsoft MAIA, a Google/Marvell TPU walk-back, or hyperscalers guiding FY27 capex flat-to-down on the late-July prints.

Thesis status

Invalidated resolved published trigger fired How this is scored →

Latest analysis and events for MRVL —

As of 2026-07-18, orbyd's latest analysis for Marvell Technology, Inc. (MRVL): Custom-silicon ASIC royalty story intact, but the AI-semi theme is mid-correction chips posted their worst month vs software on record (7/17) with money rotating into energy. Buying a falling sector with no confirmed higher low is the trap; wait for the the published invalidation level-255 May shelf to hold before re-engaging.

Invalidation trigger: a daily close that breaches the risk threshold loses the May breakout shelf and rising weekly 20-EMA. Secondary breaks: Broadcom disclosing incremental custom-ASIC share at AWS Trainium or Microsoft MAIA, a Google/Marvell TPU walk-back, or hyperscalers guiding FY27 capex flat-to-down on the late-July prints.

Next dated event on file: — catalyst in 10d.

Current Thesis

The narrative on the table is still the custom-silicon ASIC royalty stack Marvell co-designs accelerators for AWS Trainium and Microsoft MAIA (confirmed) plus a reported Google TPU program (4/20), earning recurring content per AI rack. What changed since late June is the regime, not the story. Semiconductors just posted their worst month since 2008 and their worst month ever versus software (7/17), with money rotating out of high-flying AI chips into energy (7/13, MRVL -6%). The theme has flipped from the ACCELERATING tape that carried the S&P 500 inclusion bid (effective 6/22) into a violent MATURING/correction phase. Analysts are still raising targets Keybanc to $400 (7/14), RBC $360 (7/7) but buying a falling sector with no confirmed higher low is the trap the "reset" creates. The disciplined read is to let the the published invalidation level-255 breakout shelf prove it holds before re-engaging.

Bullish and bearish views on Marvell Technology, Inc.

The model's bull view on Marvell Technology, Inc. (MRVL), in brief: Analyst targets still rising into the drawdown: Keybanc reiterated Overweight and lifted its target to $400 (7/14); RBC reiterated Outperform at $360 (7/7); Stifel $350 (6/24) and BofA $365 (6/23) both re-rated from the ~$140 spring band. The bear view: Sector in its worst drawdown in 18 years: "worst month since 2008" and "worst month ever versus software" (7/17), with a chip rout dragging the Nasdaq 100 lower (7/7). Both cases follow in full.

Bull Case

  • Analyst targets still rising into the drawdown: Keybanc reiterated Overweight and lifted its target to $400 (7/14); RBC reiterated Outperform at $360 (7/7); Stifel $350 (6/24) and BofA $365 (6/23) both re-rated from the ~$140 spring band.
  • "Buy the reset" framing emerging: 7/17 coverage argued the worst chip month since 2008 is a reset to buy, and the 7/6 "bear trap" piece framed the selloff as a shakeout inside a $1.5T AI buildout.
  • Three-logo royalty base intact: AWS Trainium + Microsoft MAIA confirmed, plus the reported Google TPU co-development (4/20) the per-rack content base is widening toward three hyperscaler accounts.
  • Structural index ownership: S&P 500 inclusion effective 6/22 installed a permanent passive holder base, reducing the float available to momentum sellers even after the one-time rebalance flow cleared.
  • Cooling macro: producer inflation plunged (7/15), pulling rates lower a tailwind for long-duration high-multiple semis, and MRVL was among names bid on the print.
  • Two-way tape, not one-way liquidation: gains on 7/6, 7/9, 7/14 and 7/16 against drops on 7/7 and 7/13 show buyers defending rather than a disorderly unwind.

Bear Case

  • Sector in its worst drawdown in 18 years: "worst month since 2008" and "worst month ever versus software" (7/17), with a chip rout dragging the Nasdaq 100 lower (7/7). High-beta AI semis take the most damage in a risk-off rotation.
  • Leadership is rotating into energy: MRVL fell 6% on 7/13 specifically as Wall Street dumped high-flyers for energy a rotation that pressures the name until it reverses.
  • New demand overhang: New York became the first state to impose a one-year data-center moratorium (7/14). If other states follow, the AI data-center buildout that underwrites hyperscaler ASIC orders faces a permitting headwind.
  • Catalyst gap: the next MRVL-specific fundamental checkpoint (Q2 FY27, ~2026-08-28 est.) sits outside the 30-day window, leaving the name to trade on sector sentiment and hyperscaler capex prints it doesn't control.
  • Broadcom is the share leader: AVGO owns the bulk of the custom-ASIC market; any disclosure of incremental volume at AWS Trainium or Microsoft MAIA resets the royalty math quickly, with ~40% of data-center revenue concentrated in 2-3 accounts.
  • Google leg is still talk: the reported TPU co-development (4/20) remains talks, not signed volume; a walk-back removes the third logo the widening thesis leans on.

Setup & Price Structure

Price ran from the ~$140 spring band to record highs into the 6/22 index inclusion, then rolled over with the broad semiconductor complex through July. The reference structure is the May breakout shelf at the published invalidation level-255, coincident with the rising weekly 20-EMA the pivot that has to hold for the uptrend to stay intact. The tape is choppy and two-sided (gains 7/6, 7/9, 7/14, 7/16; drops 7/7, 7/13), which reads as a name still searching for a base. Overhead sits a stack of $350-$400 analyst targets; below is the shelf. Chasing strength here without a confirmed higher low invites a mean-reversion flush; the higher-probability engagement is a hold-and-turn off the shelf or a reclaim of the prior highs on volume.

Catalyst Calendar (next 30 days)

  • ~2026-07-29 (est.) Microsoft and Alphabet fiscal-quarter prints; the FY27 capex guides are the most direct read on Trainium/MAIA/TPU spend and the largest MRVL tape risk in the window, both ways.
  • ~2026-07-30 (est.) Meta print; another hyperscaler capex tell for the AI-infrastructure order book.
  • ~2026-07-31 (est.) Amazon print; AWS capex commentary bears directly on the Trainium program.
  • Ongoing New York data-center moratorium (signed 7/14); watch for copycat state action as a demand-side overhang.
  • ~2026-08-28 (est.) Marvell Q2 FY27 print; just outside the window, but the pre-print blackout begins in the final days of August.

What Would Change Our Mind

a daily close that breaches the risk threshold loses the May breakout shelf and the rising weekly 20-EMA, turning the structure from pullback-in-uptrend to broken. Secondary breaks that would confirm a thesis change: Broadcom disclosing incremental custom-ASIC share at AWS Trainium or Microsoft MAIA; a public walk-back of the Google/Marvell TPU talks; or any of the late-July hyperscalers guiding FY27 capex flat-to-down. On the constructive side, a reclaim and weekly hold above the prior highs, or capex prints that raise FY27 spend, would resolve the July rout as the bear trap the bulls are calling for and re-open the accelerating leg.

Correlation Notes

MRVL trades as a high-beta member of the AI-semi basket tightly correlated to AVGO, NVDA and MU, and to the SMH/SOXX complex; the 7/7 chip rout and 7/17 "worst month" prints hit the whole group. The active cross-current is the semis-to-energy rotation (7/13), which pressures the name independent of company news. Fundamentally it is levered to hyperscaler capex (MSFT/GOOGL/AMZN/META), so their late-July guides move it more than any single MRVL headline. Micron's report is a sentiment tell for the memory/AI-semi tape. On macro, the name is rate-sensitive the 7/15 PPI plunge that cooled rates was a tailwind and it now carries two-way exposure to the data-center power/permitting theme via the New York moratorium.

Notes

  • Earnings blackout: avoid/TRIM 3 trading days ahead of ~2026-05-28 Q1 FY27 print binary risk.
  • Google-MRVL TPU partnership (2026-04-20) is the single most important datapoint on the tape; any walk-back headline = immediate exit.
  • Watch hyperscaler capex prints late-April (MSFT/GOOGL 04-29
  • META 04-30
  • AMZN ~05-01) direct MRVL tape risk both ways.
  • Cantor at $120 Neutral is the laggard catalyst alert if they blink to OW before the print.
  • AVGO pre-announcing ASIC share gains at AWS Trainium or MSFT MAIA is the single hard-stop headline breaks the thesis.
  • S&P 500 inclusion effective 2026-06-22 (announced 2026-06-05, with Flex) mechanical passive bid; rebalance trade typically executes at the 06-19 Friday close. Watch for sell-the-news reversal after the effective date.
  • AVGO pre-announcing or disclosing incremental custom-ASIC share at AWS Trainium or MSFT MAIA is the single hard-stop headline breaks the royalty thesis regardless of price.
  • Google/Marvell TPU co-development (reported 2026-04-20) is still 'talks,' not signed volume any walk-back removes the third hyperscaler logo.
  • Saturation watch: 'AI bubble' (2026-06-06) and 'is the rally over' (2026-06-04) coverage plus two-day fade = theme drifting toward MATURING; size with that in mind, not as pure ACCELERATING.
  • Next hard-fundamental checkpoint is Q2 FY27 print ~2026-08-28 (est.) outside the 30-day window; ~40% of data-center revenue concentrated in 2-3 hyperscaler ASIC accounts.
  • Teralynx T100 switch launch chatter (2026-06-03) is unconfirmed needs a real analyst note to harden the networking/optical leg.
  • binary risk.
  • Google/Marvell TPU co-development (reported 2026-04-20) is still talks, not signed volume any walk-back headline removes the third hyperscaler logo.
  • Broadcom disclosing incremental custom-ASIC share at AWS Trainium or Microsoft MAIA is the single hard-stop headline breaks the royalty math regardless of price.
  • Saturation watch: tech hit a record 39% of S&P 500 beyond the dot-com peak (2026-06-28) and sell-side PTs caught up to spot ($350 Stifel / $365 BofA) theme has drifted ACCELERATING toward MATURING; favor a pullback entry, not a breakout chase.
  • S&P 500 inclusion effective 2026-06-22 has elapsed passive demand was one-time; watch for a sell-the-news reversal in the weeks after.
  • Qualcomm entering custom chip design services (ByteDance, reported 2026-06-24) is a new entrant to the custom-silicon space monitor as a competitive datapoint.
  • ~40% of data-center revenue concentrated in 2-3 hyperscaler ASIC accounts single-customer disclosure risk is asymmetric.
  • Hyperscaler capex prints land late July (MSFT/GOOGL/META/AMZN, ~2026-07-22 to 07-31) direct read-through to ASIC demand both ways, outside any MRVL-specific binary.
  • Theme drifting SATURATED: chip stocks' worst month vs software on record (7/17), rotation into energy (7/13) size as MATURING/correcting, not ACCELERATING.
  • Hyperscaler capex prints late July (MSFT/GOOGL ~7/29, META ~7/30, AMZN ~7/31 est.) = the direct MRVL tape catalyst in-window, both ways.
  • binary risk.
  • AVGO disclosing incremental custom-ASIC share at AWS Trainium or MSFT MAIA is the single hard-stop headline breaks the royalty thesis regardless of price.
  • Google/Marvell TPU co-development (reported 2026-04-20) is still talks, not signed volume any walk-back removes the third hyperscaler logo.
  • NY data-center moratorium signed 7/14 (first state, one-year block) watch for copycat states as a demand-side overhang on the AI buildout.
  • ~40% of data-center revenue concentrated in 2-3 hyperscaler ASIC accounts concentration risk cuts both ways.
  • the published invalidation level-255 = May breakout shelf + rising weekly 20-EMA; the structural pivot to watch. Overhead target stack $350-$400 (Keybanc $400 7/14, BofA $365, RBC $360, Stifel $350).

Related · shared themes

See also · stocks to watch