Marvell AI Forecast: -3.0% Plunge but 2028 AI Boom?

How does Marvell AI Forecast fit the hyperscaler boom?

The Marvell AI Forecast is built around surging demand from U.S. hyperscalers that need to connect thousands of accelerators inside AI clusters. Marvell is a leading provider of high‑speed networking silicon, including optical and Ethernet interconnects, that maximize utilization of expensive AI GPUs from players like NVIDIA. Fast, low‑latency links are becoming mission‑critical as training models grow and inference moves into real‑time applications.

Management now guides for about $11 billion in revenue this fiscal year, implying roughly 34% growth, primarily driven by its data center business. Within that segment, Marvell expects continued strength in interconnect chips as hyperscalers expand AI clusters and upgrade to faster PCIe and CXL standards.

At the same time, the stock trades around 24 times Wall Street’s non‑GAAP EPS expectations for the current year, a multiple that looks modest versus many AI beneficiaries on the NASDAQ. That valuation is central to the Marvell AI Forecast narrative that the company offers high growth with a still‑reasonable price tag for long‑term portfolios.

Is Marvell Technology, Inc. still an AI XPU contender?

Beyond networking, Marvell’s custom AI accelerator, or XPU, business has become a major driver of its outlook. The company plays a significant role in custom silicon for hyperscalers, with design wins behind Microsoft’s Maia AI chips and Amazon’s Trainium and Inferentia lines. Concerns flared last year after reports indicated Microsoft was engaging another chipmaker for Maia 300, and Amazon was exploring alternative designs for the next generation of Trainium.

On its latest earnings call, management moved to calm those fears. CEO Matt Murphy highlighted that Marvell has purchase orders covering the entire current fiscal year for a key next‑generation XPU program and expects growth from that program to continue into fiscal 2028. He also emphasized active work on follow‑on generations, signaling that Marvell remains deeply embedded with at least one top cloud customer, widely believed to be Amazon rather than consumer‑facing giants like Apple or Tesla.

For investors parsing the Marvell AI Forecast, the custom silicon unit is crucial. Marvell expects revenue from this business to double year over year in fiscal 2027, driven by both main XPU chips and companion devices for networking, memory expansion and storage security. Some independent equity research suggests custom silicon revenue could rise from roughly $1.5 billion in FY2026 to potentially double again by FY2028 as new hyperscaler programs ramp.

What does Marvell AI Forecast imply for data center products?

The company is also expanding its role in the AI memory and interconnect stack. This week at OFC 2026, Marvell launched the Structera S 30260 CXL switch, designed to enable rack‑level memory pooling and help data centers break through the so‑called “AI memory wall”. In parallel, it introduced the Structera S 60260, billed as the industry’s first 260‑lane PCIe 6.0 switch, extending its end‑to‑end PCIe portfolio for AI scale‑up infrastructure.

These products, together with Marvell’s Alaska PCIe/CXL retimers and optical connectivity solutions, are aimed at letting data centers reconfigure bandwidth and memory more flexibly around AI accelerators. The company is even collaborating with Lumentum on optical circuit switching demonstrations, underscoring how important high‑bandwidth fabrics are becoming as models grow more complex.

Under the Marvell AI Forecast, management sees total revenue reaching around $15 billion in fiscal 2028, implying a mid‑30% compound annual growth rate from today’s base. Non‑GAAP EPS is expected to exceed $5 in that time frame, suggesting margin expansion as custom silicon and high‑value interconnect products scale.

How is Wall Street reading Marvell AI Forecast?

Institutional interest in MRVL has been robust. Large asset managers such as Vanguard, State Street, Franklin Resources and others have increased positions, while some funds have trimmed holdings after strong gains. Recent filings show overall institutional ownership above 80%, a level that keeps Marvell firmly in the conversation among S&P 500 and NASDAQ‑focused investors seeking AI exposure beyond the megacap leaders.

On the earnings front, Marvell recently reported quarterly revenue of around $2.22 billion and non‑GAAP EPS of $0.80, both ahead of consensus estimates. Analyst sentiment remains broadly positive, with a “Moderate Buy” consensus and an average price target in the low‑to‑mid $110s, implying notable upside from the current $88 range. While firms such as Goldman Sachs, Morgan Stanley and Citigroup have highlighted Marvell’s leverage to AI infrastructure, some also caution about valuation risk if the Marvell AI Forecast fails to materialize as planned.

For U.S. investors comparing opportunities across AI hardware, Marvell offers a differentiated angle versus GPU‑centric names like NVIDIA. Its focus on networking, custom XPUs and memory fabrics provides targeted exposure to the AI data center backbone rather than end‑device markets.