The number attached to Marvell’s expanded relationship with Google is $120 billion. It is large enough to invite a simple reading: a chip supplier has secured an extraordinary block of future
The number attached to Marvell’s expanded relationship with Google is $120 billion. It is large enough to invite a simple reading: a chip supplier has secured an extraordinary block of future sales from one of the largest spenders in artificial intelligence infrastructure.
That is not what the disclosed agreement says. The figure is the revenue level required for Google to earn the bulk of an equity warrant over time, rather than a stated purchase commitment or a disclosed backlog figure. The distinction matters because Marvell has said the more meaningful contribution from the relationship is expected to begin only in fiscal 2029.
For a company projecting about $18 billion in fiscal-2028 revenue, the agreement creates an unusually wide gap between the scale suggested by the long-term headline and the sales contribution embedded in nearer-term forecasts. It also puts pressure on the less glamorous parts of the investment case: product mix, gross margins and exposure to a small group of large customers.
Marvell’s filing grants Google a warrant to buy up to 58,970,907 Marvell shares at an exercise price of $206.58 per share. Of that amount, 57.61 million shares vest in 240 tranches. Google earns one tranche for each $500 million of qualifying custom-product revenue accumulated through fiscal 2033, according to the company’s Form 8-K.
Multiplying 240 tranches by $500 million produces the $120 billion figure. But the structure does not say Google is obliged to buy $120 billion of products, nor does it establish that Marvell has already booked that amount as backlog. It sets the commercial performance threshold at which the warrant shares vest.
That mechanism gives the number a different analytical role. It shows the potential scale of qualifying revenue envisioned over the agreement’s measurement period, while leaving the pace and ultimate amount of purchases dependent on the products that qualify and the revenue actually generated through fiscal 2033.
The warrant is still consequential. An arrangement that conditions a substantial equity incentive on successive revenue thresholds aligns Google’s potential ownership interest with the expansion of a custom-product relationship. Yet that alignment should not be confused with contracted sales. The filing describes a route to vesting, not a disclosed schedule of committed annual procurement.
Spread across the measurement period, the $120 billion threshold equates to roughly $18.5 billion of average annual qualifying-product revenue, according to Reuters coverage published by MarketScreener. That is comparable with Marvell’s own projected total company revenue of about $18 billion in fiscal 2028.
The comparison is not a forecast that Google will deliver an $18.5 billion annual run rate. The threshold is measured over a multi-year period, while Marvell’s fiscal-2028 figure is a companywide annual outlook. Still, placing the two numbers side by side clarifies why the warrant has attracted attention: at full scale, the qualifying revenue contemplated by the vesting design is enormous relative to Marvell’s current planning base.
Marvell raised its fiscal-2027 revenue outlook to about $12 billion, representing growth of about 45%, and lifted its fiscal-2028 outlook to about $18 billion from a prior target of roughly $16.5 billion. Those targets indicate that management already expects a rapid expansion in the business. They do not, however, answer how much of that growth comes from Google.
The company did not disclose a standalone Google revenue forecast in the cited results coverage. That leaves a central valuation question unresolved: whether the market should regard the agreement as an extension of an already rising data-centre trajectory or as a distinct earnings engine whose major effect lies beyond the periods investors are currently modelling.
Marvell Chief Executive Officer Matt Murphy said revenue from Google-related programmes through fiscal 2028 had already been reflected in prior forecasts. The deal is expected to contribute much more significantly beginning in fiscal 2029, Reuters reported.
That timing means the agreement cannot, on its own, explain the upgrades to fiscal-2027 and fiscal-2028 guidance. Google-related business remains part of the picture through those years, but the disclosed framing distinguishes programmes already incorporated from a larger later ramp. The $120 billion threshold therefore extends much further into the future than the headline itself may suggest.
There is no need to attribute all of Marvell’s present momentum to the delayed Google opportunity. In its second quarter of fiscal 2027, Marvell reported revenue of $2.739 billion, up 36.5% year on year. Data-centre revenue rose 46%, according to its quarterly filing.
Those results show a data-centre business already expanding strongly before the larger Google contribution is expected to register. They also make the timing issue sharper. Near-term growth can be real and substantial without proving the long-duration economics implied by the warrant’s maximum revenue thresholds.
Fiscal 2029 becomes the pivotal dividing line in the narrative. Until then, investors can assess Marvell largely against its stated companywide targets and the revenue already included in them. Beyond then, the question shifts to whether the custom-product programmes can scale sufficiently to make the warrant structure more than a distant ceiling.
The scope of the relationship is broader than a core tensor processing unit. Marvell identified AI inference accelerators, storage controllers, network-interface controllers, memory-interface controllers and near-memory compute among the custom-silicon areas covered by the arrangement. That range expands the possible revenue pool, since the relationship can encompass multiple components around AI systems rather than a single chip category.
It also makes execution more demanding. Delivering across several custom product lines involves converting a broad technical relationship into qualifying revenue at a pace sufficient to clear repeated vesting thresholds. The agreement’s breadth is the source of its upside, but it is also why a headline revenue figure cannot substitute for evidence of product ramps.
Marvell’s second-quarter fiscal-2027 gross margin rose to 53.1% from 50.4% a year earlier. But revenue growth is not the only variable: the company has said its ASIC end-to-end business model tends to carry lower gross margins, so a faster shift toward custom products could dilute margins even if the top line rises.
This does not make custom silicon a negative for earnings. It establishes a trade-off that the headline figure obscures: the largest prospective source of qualifying revenue may have different margin characteristics from the business mix that produced the reported quarterly improvement. As Google-related programmes become more material, revenue growth and gross-margin progression may not move together.
Customer concentration raises the stakes further. Marvell’s 10 largest customers accounted for 82% of fiscal-2026 revenue, while two customers each represented at least 10%, the company disclosed in its quarterly filing. A deeper Google relationship could reinforce strategic relevance in AI infrastructure, but it would also make execution with major customers more consequential for both growth and downside.
The test is consequently not whether $120 billion is a large number. It is whether qualifying custom-product revenue can build after fiscal 2028 at a scale that supports the warrant milestones without turning Marvell’s expansion into a lower-margin, more concentrated revenue base.
Disclaimer: This article is provided for informational purposes only. It is not offered or intended to be used as legal, tax, investment, financial, or other advice.
