Marvell Thinks AI Will Bring in a Lot More Money by 2028

Marvell Technology raised its forecast for 2028 revenue on October 6, 2026, at its Investor Day in New York City. Chairman and CEO Matt Murphy shared the update with investors, according to Reuters.

Marvell had said in advance the event would be livestreamed starting at 9 AM ET from New York. The plan was to cover company strategy, growth chances, and Marvell's part in building AI and data center equipment, according to the company's investor relations site.

The higher forecast came after a strong second quarter. Net revenue, or total sales, was $2.739 billion, a new record and up 37% from a year earlier. Marvell reported that on August 27, 2026, according to Marvell.

Gross margin, or cents left from each dollar of sales after making the products, was 53.1% under normal accounting rules, called GAAP, and 58.9% after adjustments, called non-GAAP. Adjusted earnings were $0.94 per share for the fiscal second quarter, as reported in the earnings-call transcript. That transcript also said Marvell beat Wall Street expectations for earnings and revenue.

For the full fiscal 2026, Marvell reported net revenue of $8.195 billion, also a new record. Profit under normal accounting rules was $2.670 billion, or $3.07 per share, as reported on March 5, 2026.

Even with that beat, investors sold the stock in August. Shares fell more than 8% to $221.6 in early trading on August 28, 2026, according to Reuters.

The broader context here is the gap between one quarter and a two-year plan. A 2028 target is tied to custom chip orders, network gear sales, and how much big data-center customers will spend over two full fiscal cycles. For savers, timing matters as much as size.

In my view, three things matter for reading this update. First is the gap between the two profit numbers, which means items like employee stock pay can shift reported profit even when sales grow. Second is lumpy timing. Big AI orders can move from quarter to quarter, and the August drop showed how a beat plus timing questions still pushed the stock down. Third is the long horizon. A two-year chip forecast is like a long-range weather forecast, with wide room for error when budgets, power limits and technology changes interact. What counts is whether designs turn into sales, sales turn into cash, and cash covers costs.