Yao Ren-hsun today (2nd) made a surprise appearance on stage at Computex Taipei in Nangang, Taipei, delivering a speech that immediately called Marvell the "next trillion-dollar company." MRVL stock surged over 23% pre-market to $269, hitting a new all-time high, with a market capitalization surpassing $230 billion.
(Background: Huang Ren-hsun reveals "Token Economics" at GTC 2026: Computing equals revenue, Nvidia fully mass-produces Vera Rubin)
(Additional context: COMPLETE guide to COMPUTEX 2026: Huang Ren-hsun's GTC must-see, Five major AI highlights)

Key Summary

Huang Ren-hsun at Computex declared Marvell as "the next trillion-dollar company," with MRVL pre-market soaring 23% to a record high of $269
Marvell holds 18 custom AI chip projects, with Amazon, Google, Microsoft, and Meta as clients
NVIDIA invested $2 billion in March, Q1 FY27 revenue hit a record $2.42B, and Q2 outlook raised to $2.7 billion

"Ladies and gentlemen, the next trillion-dollar company." On June 2nd, Huang Ren-hsun took the stage unexpectedly at Computex 2026 in Taipei Nangang Exhibition Center, sparking a standing ovation. He was not on his own NVIDIA podium but on the stage of Marvell CEO Matt Murphy’s keynote titled "The Future of AI Scaling Depends on Connectivity."

MRVL's stock price jumped to $269 in pre-market trading, up over 23%, setting a new record. Based on approximately 873 million shares outstanding, the market cap exceeded $230 billion.

Connectivity Becomes the Next Bottleneck for AI

In a one-hour keynote, Murphy stated: "Computing at this scale is fundamentally a connectivity challenge." His core argument is that the bottleneck in AI infrastructure is shifting from computing power to interconnectivity, and Marvell is precisely at this turning point.

Over the past decade, Marvell has invested about $36 billion to build data center platforms, with revenue more than quadrupling, and data center business accounting for 76% of total revenue. Currently, they have 18 custom AI silicon (XPU) projects, 12 of which are from Amazon, Google, Microsoft, and Meta, with 6 emerging AI clients. Revenue from custom chips is projected to reach $1.5 billion in FY2026, with a forecast to double by FY2028.

The Teralynx T100, announced early on June 1st, is also a highlight. It is the industry’s first 102.4 Tbps switch chip, built on 3nm process, with power consumption below 1,000 watts (about 25% less than competitors), supporting up to 512 ports, and samples will start shipping this quarter. Murphy emphasized that 200G per lane is the last generation for copper wiring, with 400G per lane relying entirely on optical technology.

Frequently Asked Questions

How far is Marvell from a trillion-dollar market cap?

Pre-market market cap is about $230 billion, needing roughly 4.3 times growth to reach the trillion-dollar mark. FY2027 revenue target is $11.5 billion (about 40% annual growth), but with a trailing P/E ratio of around 75, the valuation is already quite high.

Why did NVIDIA invest $2 billion in Marvell?

Through the NVLink Fusion ecosystem, NVIDIA enables cloud clients to deploy Marvell’s custom AI chips alongside NVIDIA infrastructure in the same rack, targeting the next stage of AI data center interconnectivity.

MRVL29.95%

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.