Marvell Technology Targets $90 Billion by 2031 as AI Infrastructure Demand Surges

Marvell Technology Raises Revenue Target to $90 Billion

Marvell Technology lifted its long-term sales outlook at its annual investor day, projecting fiscal 2031 revenue between $70 billion to $90 billion2

2

. The chip designer also increased its fiscal 2028 revenue outlook to $20 billion from $18 billion previously1

1

. These forecasts significantly exceed Visible Alpha consensus projections of $18.19 billion in revenue next year and $46.85 billion in fiscal 20311

1

. Marvell's stock surged 6% following the announcement, ranking among top performers in the S&P 500 and Nasdaq1

1

. Shares have more than tripled in value since the start of the year, driven by strong AI demand from hyperscalers and cloud providers1

1

.

Custom AI Chip Business Targets $30 Billion Opportunity

Marvell's custom AI chip business represents a critical growth driver, with the company now expecting revenue from this segment to exceed $12 billion in fiscal 2029 and reach approximately $30 billion by fiscal 20313

3

. The custom business generated about $1.5 billion last year and already ships custom silicon to all four major U.S. hyperscalers3

3

. CEO Matt Murphy described the progression as hyperscalers starting with Marvell's standard products, moving toward customized systems, and eventually commissioning fully custom silicon incorporating their own intellectual property3

3

. This positions Marvell Technology to capitalize on the shift toward customized AI infrastructure as data centers seek more control over their AI systems. The company's relationship with Google exemplifies this opportunity, with a broad agreement disclosed in August covering custom products tied to Google's TPU ecosystem that could generate up to roughly $120 billion in revenue through fiscal 2033 if performance targets are met3

3

.

New Connectivity Chips for AI Infrastructure

Marvell Technology launched two connectivity chips for AI infrastructure: the Alaska E 3.2T protocol-aware retimer and second-generation Alaska A 1.6T AEC DSP4

4

. The Alaska E 3.2T retimer provides 3.2 terabits per second aggregate bandwidth through 16 lanes of 224G PAM4 connectivity, supporting Ethernet, UALink and ESUN protocols with insertion loss compensation exceeding 45 decibels4

4

. The second-generation Alaska A 1.6T AEC DSP, built on 3-nanometer technology, delivers eight lanes of 224G PAM4 connectivity for active electrical cables while consuming 12.8 watts and supporting copper reaches exceeding six meters4

4

. Both products are integrated into Marvell's RELIANT interconnect telemetry platform and are currently sampling to customers4

4

. According to Vladimir Kozlov, founder and CEO of LightCounting, the AI scale-up networking market will exceed $50 billion by 2030, driven by the transition to rack-scale architectures and the move to 224G signaling4

4

.

Structera X Products Expand AI Memory Capabilities

Marvell introduced its second-generation Structera X products designed to help cloud and enterprise customers run larger AI workloads more efficiently2

2

. The products increase available memory, improve how systems share that memory, and help customers get more use from existing hardware2

2

. Marvell said the new products can deliver up to 40%-50% faster response times for large AI models or support roughly twice as much context at similar speeds compared with the previous generation2

2

. The company is working with AMD, Arm, Alibaba, Samsung Electronics and SK Hynix to support broader adoption2

2

. Marvell expects the new Structera X products to begin sampling in the first quarter of 20272

2

.

Strategic Position in Evolving AI Ecosystem

Marvell's growth strategy positions it alongside Nvidia rather than in direct competition. The companies announced a strategic partnership in March under Nvidia's NVLink Fusion platform, allowing Marvell to provide custom XPUs and networking compatible with Nvidia's architecture3

3

. Nvidia also invested $2 billion in Marvell3

3

. BNP Paribas analyst Karl Ackerman noted that Marvell expanded its estimated total addressable market to $400 billion by 2030, including about $385 billion tied to data centers2

2

. Within that market, Marvell estimates a $235 billion opportunity for custom products, $85 billion for switching and storage, and $65 billion for interconnect products2

2

. Ackerman projects Marvell's updated forecasts could support earnings of about $33.40 per share by 2030, above management's target of more than $30 and the current consensus estimate of $24.912

2

. Marvell plans to showcase its AI data-center technologies at the 2026 OCP Global Summit in San Jose from Oct. 12-15, highlighting networking, memory, storage, connectivity and security products designed to improve AI infrastructure performance and efficiency2

2

.