Marvell up pre-market on rumors about winning Google’s TPU/MPUs custom silicon programs. 🚀

Marvell quietly turned out to be one of our strongest wins in the TPO Portfolio, gaining 65% 💪 since we initiated our position back in November 2025 and kept expanding it to a near “full” allocation by mid March.

🏆 So, how did we manage to get Marvell so right? (Even though the waiting period between November-March was quite painful at times)

We got sold on Matt Murphy’s strategic pivot to Optics and Memory at a conference last year. Our opinion at the time as simple as we doodled it out below 👇

1️⃣ The whole market was focused on Marvell’s ASIC business. AVGO was winning entire customers from the top, while Taiwanese players were picking off customer programs from the bottom leaving Marvell stuck in between.

2️⃣Marvell kept its Trainium program commitments it had won on track but began the silent pivot to connectivity, specifically towards Optics and Memory Management

3️⃣The plan Murphy laid out at the conference was simple: investors should actually focus on Optics and their Growth segment in their DC biz (mem controllers, switches) which was projected to grow at a 40% CAGR, way faster than XPUs growth rates.

4️⃣PLUS PLUS PLUS! Optics and Memory connectivity businesses would fetch higher margins since Marvell could leverage their own IP to work with partners

5️⃣2 months later, in December, Marvell bought Celestial AI and XConn further deepening our conviction and we bought more shares of Marvell to our TPO portfolio.

Bottom Line: 6 months later, we are glad we stayed true to our fundamental conviction, especially as we chose to invest in Marvell, instead of Broadcom at the time, with the former returning close to 60%, vs. the latter at 16% during this period of time.

Join us at The Pragmatic Optimist and get full access to our deep dive research and our TPO Portfolio (with Live Trades), which is up 14.11% since March 1 📈, hugely outperforming the indices and benchmark ETFs.