$MRVL
MediumInvestment Thesis
Marvell's Celestial CPO architecture positions them as the 'mini-Broadcom' of AI networking ASICs. MSFT Maia ASIC ramp H2 2026 + Google TPU development talks for 2028 provide multi-year revenue visibility. The Celestial acquisition was transformative. Post-POET NDA breach, MRVL is likely sourcing lasers directly from SIVE — actually improving Sivers' position.
Catalysts
- MSFT Maia ASIC ramp H2 2026
- Google TPU ASIC development talks — two chip models
- Amazon photonic fabric purchase agreements + warrants
- Celestial CPO architecture now multi-sourcing lasers directly from SIVE
Risks
- Custom ASIC concentration — few customers drive most revenue
- NVDA custom silicon competition intensifying
From the Feed (6 signals)
All the hyperscalers $SIVE likely ends up in 2027-2028 is staggering at a $900m MC. Markets don't understand what's coming. From speculative mapping: SIVE → POET → MRVL → AMZN/MSFT/GOOGL. SIVE → JBL 1.6T → META/NVDA. SIVE → Ayar → AlChip/GUC → AMZN/AMD. SIVE → O-Net ELS → Asian hyperscalers (Tencent/Alibaba/ByteDance). SIVE → AAPL Silicon Photonics (50M units/yr). Alpha comes from future revenue proportional to demand from every Western/Asian hyperscaler for CPO/1.6T in 2027, 2028, 2029, and onward.
Did you listen anon? The fact that $SIVE is up 600%+. But still can 10x from here in a year... once ~ $AAPL, $JBL, and $MRVL require mass production of their lasers in 2027. Is incredible. Probably my most legendary thesis post since $AXTI.
Just putting out there... Would have been +15.02% in 2W equal-weighted return on 30 different stocks. $INTC +29.62%, $MRVL +40.95%, $TSM +4.72%, $COHR +18.9%, $RKLB +26.76%, $DRAM +12.29%, $AVBO +18.32%, $AMZN +9.17%, $ARM +36.6%, $NBIS +15.22%, $GOOGL +6.41%, $AMKR +32.25%, $HOOD +19.14%, $HIMS +42.53%, $SMTC +18.83%, $POWL +9.26%, $MSFT +11.44%...
Frontrunning 1.6T/CPO within the broader photonics supercycle is the most compelling investment to me. My high conviction: 1. $SIVE - laser revenue scales aggressively with $JBL, $MRVL, Ayar, O-Net. 2. Shunsin - Foxconn's optical foundry at $1.5B MC vs LWLG. 3. Win Semi - foundry for Sivers + AVBO + SpaceX supply chains. 4. $MRVL - mini-Broadcomm. 5. $HPS.A - Transformers. There's many others like $NBIS, $JBL, $RPI, $TSEM, $LITE, $ARM, $SOI, $AXTI, $IQE, $ALRIB that I'm very fond of.
$MRVL cancelled $POET purchase orders after the CFO went out and violated NDA when getting angry. Ouch to Poet, down -46%, this is why I don't like companies with single customer concentration risk. On the bright side for $POET holders they do have $420m cash buffering downside risk and a few other customers (though Marvell was basically the entire Poet bull case story). It does look like Marvell delayed their own timelines as well by doing this.
I am now long MSSCorps (6830) ~$1.4B MC. This appears to be a functional monopoly in CPO for inspection. For customers from mapping: TSM, NVDA, AAPL, AMAT, LRCX, ASML, INTC. And high probability AVBO, MediaTek, Samsung, MRVL, and others. If you're curious — they did mention EU too. For inspection (non-destructive infrared (IR) leakage detection), they're a monopoly. And have aggressively used litigation (Enli Tech lawsuit) to lock out rivals, which I view as a positive thing.