SKN  | Marvell’s $18 Billion Revenue Target Signals a Broader AI Infrastructure Opportunity

Key Points

• Citi raised its FY2027 Marvell revenue estimate to approximately $12 billion and FY2028 to $18 billion, with data center revenue expected to grow more than 60% in both years.
• The expanded Google agreement covers Marvell’s broader XPU-attach portfolio, while new CPO and NPO optical design wins could create additional long-duration growth.
• Citi maintains a Buy rating and $275 price target, but the valuation increasingly depends on Marvell converting AI infrastructure demand into sustained margins and earnings growth.

Why Citi Is Extending Marvell’s Growth Horizon

Marvell Technology is increasingly being positioned as more than an AI connectivity semiconductor company. Citi’s latest assessment suggests the company is developing into a broader custom computing and infrastructure platform, with visibility extending well beyond the current earnings cycle.

In its August 28 report, Citigroup raised its FY2027 revenue expectation from $11.5 billion to approximately $12 billion and lifted its FY2028 forecast from $16.5 billion to approximately $18 billion. Data center revenue is expected to grow by more than 60% year over year in both fiscal years.

The change is important because Citi has moved its valuation base year from 2027 to 2028. That decision reflects greater confidence in the visibility of Marvell’s two-year revenue outlook rather than simply an increase in near-term earnings expectations.

For long-term capital, the distinction matters. The investment thesis is increasingly being built around multi-year infrastructure commitments rather than a single AI spending cycle.

Google Expands the Addressable Opportunity Beyond Core Compute

One of the most consequential developments is Marvell’s expanded commercial agreement with Google.

According to Citi, the agreement covers multiple categories of XPU-attach products, including AI inference accelerators, storage controllers, network interface controllers, memory interface controllers and near-memory computing.

The distinction between XPU-attach and core compute is strategically important. Citi’s interpretation is that Marvell is not replacing Broadcom as the supplier of Google’s core TPU compute dies. Instead, Marvell is positioned across the surrounding infrastructure required to connect, control and support those computing systems.

Individually, these projects may be smaller and more fragmented than a core accelerator program. Collectively, however, they create a substantial addressable market.

Management continues to target $10 billion to $11 billion of custom chip revenue by FY2029 and expects the custom business to accelerate materially during the second half of FY2027. That timeline provides investors with a defined period in which the current pipeline should begin translating into larger commercial contributions.

Optical Networking Could Become the Next Growth Engine

Marvell’s opportunity is not limited to custom silicon.

Management described scale-out optics as one of its most compelling areas and disclosed additional CPO and NPO design wins secured over recent quarters. Citi estimates the Celestial AI CPO and photonics fabric opportunity at approximately $150 million, while the broader expansion of optical products, including NPO, could represent approximately $300 million.

The strategic significance is larger than the initial revenue figures.

As AI clusters become increasingly distributed, moving data between processors becomes a fundamental infrastructure constraint. Marvell’s simultaneous investment across CPO, NPO and scaling interconnect technologies therefore provides exposure to multiple architectures as the industry determines which approaches will dominate.

The October 6 Investor Day becomes particularly important because management is expected to provide greater visibility into the technology roadmap and potential revenue contribution.

The Earnings Model Is Beginning to Reflect Operating Leverage

Citi estimates Marvell’s total data center revenue at approximately $9.8 billion in FY2027 and $16.5 billion in FY2028.

XPU and XPU-attach-related revenue is expected to represent roughly 20% of data center revenue in FY2027, or approximately $2 billion, rising to around 28%, or approximately $4.7 billion, in FY2028.

At the same time, Citi expects non-GAAP operating expenses in FY2028 to grow at roughly half the rate of revenue. That creates the potential for meaningful operating leverage as the company scales.

Citi expects non-GAAP operating margins to reach Marvell’s long-term target range of 38% to 40% during the fourth quarter of FY2027 and move toward the upper end of that range during FY2028.

For investors, this may ultimately matter more than the revenue headline. Sustained margin expansion would allow incremental AI infrastructure revenue to translate disproportionately into earnings.

What Could Challenge the Thesis

The growth outlook remains dependent on continued strength across several semiconductor markets.

Weak storage demand, slower networking growth and loss of market share in storage or networking represent important downside risks. The company’s increasingly ambitious expectations also create execution risk: design wins must become production volumes, production volumes must generate acceptable margins, and customer concentration must not undermine the economics of growth.

There is also a timing consideration. Much of the thesis depends on projects accelerating from the second half of FY2027 onward. Investors therefore need to distinguish between announced opportunities and revenue that has already entered the company’s earnings base.

Citi’s $275 price target is based on a 28x forward P/E multiple using CY2028 earnings, consistent with Marvell’s average P/E over the preceding three years. That valuation framework assumes that the expected revenue expansion and operating leverage will materialize.

Closing Insights: Marvell Is Becoming a Longer-Duration AI Infrastructure Story

The most important development in Citi’s report is not simply that Marvell raised its revenue forecast.

It is that the company’s growth visibility is expanding across several layers of AI infrastructure simultaneously.

The Google agreement broadens the XPU-attach opportunity, optical design wins create another potential growth vector, and rising data center revenue provides the scale required for operating leverage. If those elements develop as management and Citi expect, Marvell could increasingly be valued on the durability of its AI infrastructure platform rather than on quarterly semiconductor cycles.

That also raises the standard for execution. At a $275 price target based on 2028 earnings, investors are paying for meaningful future growth and margin expansion today. The October 6 Investor Day could therefore become an important test of whether the long-term opportunity is progressing from strategic narrative into measurable commercial scale.

For sophisticated investors, the central question is no longer whether AI infrastructure spending is growing. It is whether Marvell can secure enough of that spending, sustain its competitive position and convert the resulting revenue into durable free cash flow and capital efficiency.

For a confidential discussion regarding technology-sector investment strategy, cross-border capital allocation, private-market opportunities, digital infrastructure exposure, or long-term wealth structuring around emerging technology themes, contact our senior advisory team.