📌 Marvell Data Center Acceleration and Custom Revenue Test

BullSignal Automated Editorial System Published

The focus of Marvell’s latest earnings is not merely that quarterly figures came in slightly above expectations, but that growth in its data center business accelerated significantly. The company’s second-quarter revenue was USD 2.739 billion, up 37% year over year; adjusted earnings per share were USD 0.94. Data center business grew 46% year over year, above the previous quarter’s 27%. The company’s guidance for next quarter’s revenue and earnings per share was also above prior levels, and it raised its revenue expectations for the full year and next year.

This provides short-term operating validation that AI infrastructure demand is still expanding. For Marvell, accelerating data center revenue means that orders and deliveries of related products are being converted into current-period revenue, rather than merely reflecting customers’ capital expenditure plans. Whether next quarter’s guidance can be met will be a direct test of whether this growth pace continues.

The medium-term narrative extends to inference scenarios. The company said that CXL has been deployed at scale by multiple customers. CXL can be used to expand memory and is particularly suited to AI inference’s demand for memory capacity. If deployments continue to expand, Marvell’s opportunity will come not only from traditional interconnects, but also from newly added memory expansion demand in inference architectures.

But longer-term growth remains primarily a management target. The company expects FY28 revenue of approximately USD 18 billion, with data center business growth still exceeding 60% and custom revenue more than doubling year over year. This assessment assumes that the AI project pipeline continues to convert into revenue. Current earnings and guidance have provided short-term validation; whether CXL deployments can continue to expand, and when custom projects generate revenue, will determine the credibility of the medium-term targets.

Sources