Marvell Raises 2028 Revenue Forecast to $20 Billion as AI Data-Center Demand Surges

Marvell Technology raised its fiscal 2028 revenue forecast to about $20 billion on October 6, 2026, as AI data-center demand continues to expand beyond GPUs into networking, custom silicon and high-speed connectivity. The company’s updated outlook is one of the clearest signals yet that the artificial-intelligence infrastructure boom is spreading across a much wider semiconductor supply chain.

Marvell raises its 2028 revenue forecast

Marvell’s new forecast is above its previous target of about $18 billion. The company is benefiting from cloud providers and hyperscale customers investing aggressively in infrastructure needed to train and run large AI models.

AI systems require far more than processors alone. Large clusters need advanced networking chips, optical interconnects, switching, storage connectivity and increasingly specialized silicon designed for particular workloads. Marvell participates in several of those areas, making it a useful indicator of how broad AI capital spending has become.

Why custom silicon is becoming more important

Large technology companies are increasingly designing or commissioning custom chips to complement standard accelerators. Custom silicon can improve power efficiency, lower costs and optimize performance for specific tasks. That trend creates opportunities for semiconductor companies that can help customers design and connect specialized hardware at scale.

Marvell has positioned itself as a supplier to that market while also expanding in data-center networking. As AI clusters grow larger, moving data efficiently between processors becomes a major technical and economic challenge. A bottleneck in connectivity can reduce the value of expensive computing hardware.

What the forecast says about the AI infrastructure cycle

The higher revenue target suggests Marvell expects the current data-center investment cycle to continue for several years. That view is consistent with the broader industry trend in which cloud companies are committing tens of billions of dollars to AI infrastructure, energy and data-center construction.

However, long-range forecasts carry risk. AI spending could slow if customers become more disciplined on returns, if power constraints delay new data centers, or if competition pushes prices lower. Semiconductor demand can also be cyclical, especially when customers build inventory faster than end demand grows.

Why investors are watching margins as well as growth

Fast revenue growth is important, but investors will also focus on gross margins, free cash flow and the cost of supporting custom programs. Large contracts can be attractive, yet they may require heavy engineering investment and close collaboration with a small number of powerful customers.

Concentration is another risk. The AI infrastructure market is dominated by a limited group of cloud and technology companies, so changes in their capital-spending plans can quickly affect suppliers.

What to watch next

The next important signals will be new hyperscaler contracts, customer concentration, data-center networking growth and progress in custom AI silicon. Investors should also watch whether cloud providers continue raising capital expenditures into 2027 and 2028.

Marvell’s forecast reinforces a broader theme already visible in AMD’s plan to increase 2027 chip supply: AI demand is increasingly being planned years in advance across the semiconductor ecosystem.

Source

Based on same-day reporting from Reuters.