Marvell shares jumped ~10% after the company disclosed an expanded custom chip partnership with Google, including a warrant for 58,970,907 shares at a $206.58 exercise price (about $12.2B at the strike). The warrant vests in 240 installments tied to Google’s qualifying purchases of custom chips—one tranche per $500M of revenue, with measuring periods running Aug. 1, 2026 through Jan. 29, 2033—implying a $120B purchase ceiling if fully reached. Marvell also reported record fiscal Q1 revenue of $2.4B (+28% YoY) and guided to about $2.7B in fiscal Q2 (+35% at the midpoint), while full warrant exercise would add ~7% to the current share count.
This is more a customer-concentration event than a clean revenue upgrade. The economic signal is that Google is willing to use equity-linked incentives to secure custom silicon capacity, which lowers procurement friction but also makes MRVL increasingly tethered to one buyer’s AI capex cadence. That tends to support the stock in the near term, but it also raises the quality-of-earnings discount: as the mix becomes more Google-dependent, investors usually pay a lower multiple for the same dollar of growth.
Second-order, this is modestly negative for merchant AI silicon leaders because it suggests hyperscalers are still pushing inference and networking workloads into bespoke architectures. NVDA is not directly threatened on training demand, but any durable TPU expansion is a share-transfer risk in inference-heavy spend over the next 6-18 months; the real pressure point is where custom compute, NICs, and memory-interface silicon take budget that might otherwise have gone to broader accelerator ecosystems. If the custom stack scales, the biggest beneficiaries are Google’s cost structure and latency profile, not necessarily Marvell shareholders, because the warrant hardwires dilution and gives Google upside participation at a fixed strike.
Contrarian view: the market may be capitalizing the optionality as if it were already contracted spend. The vesting only matters if Google keeps buying, and that path can be reversed quickly if AI capex slows, TPU economics underwhelm, or Marvell’s next couple of quarters fail to show margin-stable acceleration. The thesis breaks if MRVL cannot hold post-gap gains into earnings or if data-center growth decelerates while concentration risk becomes the dominant investor focus.
AI-powered research, real-time alerts, and portfolio analytics for institutional investors.
Overall Sentiment
strongly positive
Sentiment Score
0.55
Ticker Sentiment
