Marvell Technology (NASDAQ:MRVL) has suddenly become one of the most closely watched names in the custom AI-chip race. The Marvell Google AI deal could generate up to roughly $120 billion of custom-product revenue through fiscal 2033 if the required milestones are reached. Google also received warrants covering 58.97 million Marvell shares at $206.58 each. That could represent about $12.2 billion of stock if fully exercised.
The announcement sent Marvell shares sharply higher and put pressure on Broadcom, Google’s established custom-chip partner. But the headline number is only the beginning of the story. Marvell reports fiscal Q2 results on August 27, giving investors their first major financial checkpoint since the agreement was announced.
The central question is straightforward: how much of that $120 billion opportunity is actually likely to become revenue?
The Marvell Google AI Deal Is Bigger Than Marvell’s Current Base
The easiest way to understand the scale of the Google opportunity is to compare it with Marvell’s existing business.
Marvell generated a record $2.418 billion of revenue in fiscal Q1 2027, up 28% from the prior year. Data center revenue was $1.83 billion and represented 76% of company sales. Management then guided fiscal Q2 revenue to about $2.7 billion, plus or minus 5%.
Even if you annualize that record Q1, Marvell was running at less than $10 billion of yearly companywide revenue. The Marvell Google AI deal, by contrast, could eventually represent $120 billion of cumulative revenue from a single customer.
That comparison is important because $120 billion is not simply another large design win. It represents a possible change in Marvell’s scale.
Management was already forecasting nearly $11.5 billion of fiscal 2027 revenue before this announcement. It also expected fiscal 2028 revenue to reach roughly $16.5 billion. More importantly, the custom business was expected to more than double during fiscal 2028.
That creates an important question for August 27. Investors need to determine how much of Google was already reflected in those forecasts.
If much of the program was already included, the announcement may provide greater visibility rather than immediate upside. If it is substantially incremental, Marvell’s longer-term forecasts could eventually move much higher.
The Warrants Turn Google Revenue Into A Visible Scoreboard
The structure of the Google agreement makes this story unusually measurable.
Google received the right to purchase 58.97 million Marvell shares at $206.58 each. If exercised in full, that position could make Google one of Marvell’s largest shareholders. Reuters reported that vesting is connected largely to Google purchasing qualifying custom products from Marvell.
That matters because the $120 billion figure should not be confused with guaranteed backlog. It represents the potential revenue associated with reaching the agreement’s full performance thresholds.
In practical terms, investors can now watch recognized Google revenue as evidence that the Marvell Google AI deal is progressing.
That is different from a normal semiconductor design-win announcement. Design wins can take years to reach production. Customer forecasts can also change before chips reach meaningful volume.
Marvell itself made that point indirectly on its May earnings call. Management said new custom sockets normally require roughly two years before contributing meaningful revenue. It also said demand across existing programs had increased materially.
The Google warrant therefore creates alignment between the two companies. Google earns more of its potential equity position as the commercial relationship grows.
For shareholders, that provides a simple framework. The real story is not the maximum $120 billion figure. It is how quickly qualifying revenue starts accumulating toward that figure.
August 27 Must Clarify The Timing & Economics
Marvell’s August 27 earnings call arrives at exactly the right moment.
Before the Google announcement, management was already describing an aggressive custom-silicon expansion. Custom revenue was expected to grow more than 20% in fiscal 2027. It was then expected to more than double in fiscal 2028. Management also reiterated its goal of generating more than $10 billion of annual custom revenue in fiscal 2029.
Those numbers now become the baseline against which the Marvell Google AI deal must be judged.
Investors should listen for any change to fiscal 2028 or fiscal 2029 expectations. They should also watch whether management provides clues about the first meaningful Google production ramp.
Margins are equally important. Marvell reported a 58.9% non-GAAP gross margin in Q1. It guided Q2 to between 58.25% and 59.25%. Management has said product mix remains an important driver of gross margin.
A massive hyperscaler program can create enormous revenue without producing identical margins to smaller semiconductor businesses.
Capacity adds another layer. Marvell previously said it planned roughly $1 billion of supplier prepayments during fiscal 2027 to secure future materials. Management is giving key suppliers five-year forecasts and backing those forecasts with cash.
So August 27 is about more than revenue. Investors need evidence that Marvell can manufacture, finance and profitably scale the Google opportunity.
Broadcom Competition Could Decide How Incremental Google Really Is
The competitive question may be the hardest part of the story.
Broadcom has played an important role in Google’s custom AI infrastructure. The Marvell Google AI deal immediately raised the possibility that Marvell is taking business from its larger rival. Broadcom shares fell more than 5% following the announcement, while Marvell moved higher.
Still, the situation may not be that simple.
Reuters noted that analysts see the agreement as potentially reflecting Google diversifying its AI-chip supply chain, rather than completely replacing Broadcom. The partnership also reaches beyond a single processor. Marvell can participate in technology surrounding Google’s TPU systems, including processing, networking, memory and storage infrastructure.
That distinction matters for Marvell’s long-term revenue.
If Google’s custom-silicon spending pool is simply becoming much larger, both Marvell and Broadcom could continue growing. Marvell would gain incremental business without requiring major share losses by Broadcom.
If Marvell is actually displacing Broadcom in major future platforms, the strategic implications are larger.
Marvell already said it has custom engagements across every major U.S. hyperscaler. Management also described custom-design activity as exceptionally strong and said individual programs continue getting larger.
The August call may not identify customer programs by name. Still, comments about new XPUs, attach products and production schedules could reveal whether Google is becoming a much larger part of Marvell’s custom-silicon roadmap.
Final Thoughts
The Marvell Google AI deal clearly expands the possible ceiling for Marvell’s AI business. Breakingviews estimated that a successful ramp could lift projected 2032 revenue from roughly $43 billion to about $62 billion. Even that scenario, however, would leave Marvell far short of the $1 trillion valuation sometimes discussed around its long-term AI opportunity.
That makes valuation especially important after the rally.
As of August 20, Marvell was trading at roughly 25.42x LTM enterprise value to revenue, 25.25x LTM price-to-sales and 81.68x LTM enterprise value to EBITDA. Its LTM P/E stood near 86.70x, while LTM EV/EBIT was about 154.80x. Those are demanding multiples that already reflect substantial expectations for future growth.
The valuation does not mean the Google opportunity cannot justify further expansion over time. It does mean execution now carries greater weight. Revenue timing, margins, capacity requirements and competitive share gains all matter when investors are paying such high trailing multiples.
That is why August 27 is so important. The Google agreement gave investors a spectacular maximum number. The next earnings report needs to begin showing the financial path toward it.
Disclaimer: We do not hold any positions in the above stock(s). Read our full disclaimer here.
