When a Stock Doubles, “Not Until 2029” Is a Death Sentence

(SeaPRwire) –   By: Reginald Vance

The number that killed Marvell’s stock after hours was not the revenue miss. It wasn’t the EPS beat. It was one sentence delivered on an earnings call. CEO Matt Murphy confirmed that Google’s custom chip revenue won’t contribute materially to Marvell’s P&L until fiscal 2029. That’s a two-year gap. The market had ridden MRVL up 184 percent through 2026 before the report dropped. A stock that doubles in twelve months doesn’t want modest beats. It wants execution arriving this quarter. Murphy gave them everything Wall Street penciled in. Adjusted EPS hit 94 cents against a 93-cent estimate. Revenue landed at 2.74 billion versus the 2.72 billion consensus. Data center revenue exploded 46 percent year-over-year to 2.2 billion. Every single metric cleared expectations. The stock still dropped six percent in after-hours. The reason is uncomfortable for anyone who has priced momentum into a semiconductor name. When capital is this concentrated around a single narrative, speed becomes the only currency that matters. The Google warrant was worth up to 12.2 billion in equity for Alphabet. The revenue opportunity stretches to 120 billion through fiscal 2033. The timing between signing and shipping is where these deals lose credibility.

Here is what actually changed on the order book. Marvell raised fiscal 2027 guidance to roughly 12 billion, up from 11.5 billion. Fiscal 2028 jumped to 18 billion from 16.5 billion. Those are meaningful upward revisions. Q3 revenue guidance came in at 3.15 billion midpoint against a 3.04 billion consensus. EPS guidance of 1.05 to 1.15 sat above the 1.08 call. Murphy said custom-chip revenue will more than double next year. He also confirmed that existing custom targets through fiscal 2028 already reflect some Google revenue. The bulk of the deal lands in 2029 and beyond. The partnership covers AI processors, storage, networking, and memory hardware connected to Google’s tensor processing units. Bob O’Donnell at TECHnalysis Research made a useful observation about the broader backdrop. He noted that expectations around custom AI accelerator projects are riding very high, especially given recent news about the Broadcom and OpenAI Jalapeno collaboration. Marvell’s non-GAAP gross margin came in at 58.9 percent, down 50 basis points year-over-year. That compression tells its own story. Custom designs at scale eat margins before they expand them. You see this pattern every time a fabless house transitions from volume production to bespoke silicon. The foundry capacity and wafer allocations needed to support that scale of custom silicon through 2033 represent a logistics problem of unprecedented magnitude. Someone has to build the fabs or lease them at a premium.

The market is looking at the wrong number. It is fixated on the 2029 timing and missing what Marvell is already doing. Custom-chip revenue more than doubling next year does not require Google to ship. That growth is already baked into current guidance. The Google partnership layers incremental upside on top of existing custom wins. Murphy acknowledged upside bias to the prior 10 billion plus fiscal 2029 target but declined to publish a new number. Details are saved for the October 6 investor day. That date, not the earnings report, is the real catalyst. The uncomfortable truth about AI chip consolidation is this. Big Tech is racing to replace Nvidia’s expensive processors with in-house alternatives. Marvell has been a direct beneficiary of that trend. But beneficiaries are not owners. Broadcom is building its own custom AI engine. Google could eventually internalize more design work. The 12.2 billion warrant gives Alphabet a stake in Marvell’s equity and a seat at the table. It also gives them alignment through the long ramp. The endgame isn’t Marvell losing Google as a customer. It is Marvell becoming so embedded in Google’s silicon roadmap that the supplier-developer distinction dissolves. That is when you have built a real moat. That is also when your margin profile starts looking more like a foundry’s than a fabless chipmaker’s. October 6 will reveal which path Murphy is walking. Until then, a 58.9 percent gross margin and a six-percent after-hours drop is what it looks like when a company scales faster than the market can price.

Author bio: Reginald Vance, a venture partner specializing in semiconductor valuation and advanced materials, with two decades covering fabless chipmakers and custom silicon procurement cycles.