$CRDO FY27 still requires a pretty sharp second half. Q1 came in at $479M and Q2 is guided to $525–535M. At the midpoint, that puts H1 revenue at roughly $1.01B. To reach the >80% FY27 growth target, Credo would need about $1.39B in H2 revenue, or roughly $693M per quarter. That is about 31% above the Q2 midpoint. So the setup is pretty simple from here: the H2 ramp needs to be real. And that is why I care much more about what management says tonight about optical revenue and the timing of those ramps than whether Q1 beat estimates by a few million dollars.

I spent the last week rebuilding my $MRVL thesis after Q2 and the latest news with $GOOGL etc. Posting a small summary of it here, the full article can be found in bio/comments. So, the stock fell 10% after earnings, but I came away more positive on the business than I was before the report and surprisingly, Google is only part of the reason. Back in July, my problem with Marvell was fairly simple: I could see the technology platform much more clearly than the shareholder economics. Custom silicon, optics, switching, CXL, SerDes… lots of interesting pieces. What I could not prove was whether those pieces actually worked together commercially. I think we have alot more evidence now and Google is the first big one. The famous $120B figure is NOT backlog. It is the ceiling for qualifying purchases under the warrant agreement. What interests me much more is what Google is actually working with Marvell on such as accelerators, NICs, storage controllers, memory interfaces and near-memory compute. This looks alot like a platform relashionship. Then came Q2 this thursday. Revenue reached $2.74B (+37% YoY), Data Center grew 46%, FY27 revenue guidance moved to roughly $12B and FY28 to $18B. But the biggest change in my thesis is probably scale-up and connectivity. Three months ago Marvell was talking about roughly $300M of FY28 scale-up optics revenue. Now scale-up optics is a significant contributor to the entire $1.5B FY28 guidance raise, with NPO developing alongside CPO, switching expanding and multiple customer engagements emerging. Connectivity may actually be carrying more of the Marvell story than I expected. We also learned something important about Google: revenue covered by the agreement through FY28 is already included in existing custom guidance. So adding billions of “new Google revenue” to FY28 would be double counting. The interesting part starts in FY29, where management says the previous $10B+ custom silicon outlook now has clear upside. CXL also moved higher on my list. Marvell says it is already being deployed across multiple hyperscalers at very high volumes, with memory scarcity accelerating adoption. The financial side is improving too. Gross margin will fall as custom grows, but operating leverage is currently more than compensating: operating margin hit 36.6%, with management expecting 38–40% from Q4 and toward the high end in FY28. I am still watching the denominator closely. Diluted shares are now around 921M, the Google warrant can add another ~59M shares, and we still do not have enough disclosure on the warrant accounting. So my old concern has not disappeared: Revenue growth eventually has to become EPS and cash flow per share. But compared with July, I now have much more evidence that Marvell’s technology platform can turn into an economically meaningful one.