Evercore lifts Marvell target to $155 amid AI shift

Evercore lifts – Evercore ISI raised its Marvell Technology (MRVL) price target to $155 from $133 and kept an Outperform rating, pointing to a market shift from training to inference by the end of 2026. Another analyst also boosted Marvell’s outlook, while the broader debate c

When AI buyers start spending differently, the stock market takes notice. On May 19, Evercore ISI lifted its price target on Marvell Technology, Inc. (NASDAQ:MRVL) to $155 from $133 and maintained an Outperform rating.

The move followed a round of Q1 AI channel checks, during which the firm zeroed in on a theme it said is reshaping demand: AI workloads are expected to shift from a training-focused market toward an inference-led market by the end of 2026.

Evercore ISI tied that transition to practical pressures inside data centers—attention moving toward cost-per-token. return on investment. and total cost of ownership. In the firm’s view. that is pulling hyperscalers toward internally developed ASICs and alternative accelerators. categories where Marvell is positioned to benefit.

A day earlier, May 18, another Wall Street forecast added momentum. Melius Research analyst Ben Reitzes raised the firm’s price target on Marvell (MRVL) to $220 from $140 while keeping a Buy rating. Reitzes said that while “nothing really emerged as incrementally good from Trump going to China. ” the firm had become “incrementally good” on memory and AI semiconductor companies.

Melius said it raised its long-term estimates and price targets for several Buy-rated “bottleneck stocks”—Micron. Sandisk. AMD. Intel. and Marvell—and also kept Qualcomm at Hold. The firm also said it continues to believe semiconductor companies will capture market capitalization growth—or at least more upside—compared with traditional software companies and non-semiconductor members of the Mag 7 over the long term.

Marvell Technology, Inc., along with its consolidated subsidiaries, supplies data infrastructure semiconductor solutions across the data center core and network edge. The company designs, develops, and sells integrated circuits.

The debate now is less about whether AI infrastructure matters and more about who captures the most upside as spending priorities evolve. Evercore ISI’s channel-check takeaway leans on the economics of inference—cost-per-token and total cost of ownership—while Melius frames the bigger picture around semiconductors as bottlenecks and relative upside versus software.

There’s also a separate. cautionary note in the broader commentary: while some investors may see MRVL as an investment opportunity. others argue that certain AI stocks may offer greater upside potential and carry less downside risk. The same piece points readers toward a free report described as covering an “extremely undervalued AI stock” expected to benefit from Trump-era tariffs and the onshoring trend.

For now. the immediate takeaway from May 19 is straightforward: Evercore ISI is betting that the training-to-inference shift by the end of 2026 will keep demand strong. and that Marvell can ride it. The company’s next moves won’t be in headlines—they’ll be in whether the industry’s cost pressures keep steering hyperscalers toward the kinds of chips built for the next phase of AI workloads.

Marvell MRVL Evercore ISI Melius Research AI infrastructure inference training cost-per-token hyperscalers ASICs price target