On June 2, 2026, at COMPUTEX Taipei, NVIDIA CEO Jensen Huang made a surprise appearance during Marvell CEO Matt Murphy’s keynote. Addressing the audience, he declared, "The next trillion-dollar company, ladies and gentlemen."
The market reaction was nearly instantaneous. MRVL shares surged over 27% intraday and closed up 32.52%, adding roughly $62.4 billion in market value in a single day. Marvell’s market cap crossed the $250 billion mark, setting a new record for the company’s largest one-day gain.
But what is the foundation for Jensen Huang’s prediction? What objective milestones must Marvell reach to grow from a market cap of about $192 billion to $1 trillion?
Current Baseline: Data at the Time of Jensen Huang’s Endorsement
Before Jensen Huang publicly backed Marvell, the company had just released an earnings report that exceeded market expectations.
For FY2027 Q1 (ended May 2, 2026), Marvell posted net revenue of $2.418 billion, up 28% year-over-year and 9% quarter-over-quarter, beating the company’s previous guidance midpoint by $180 million and setting a new single-quarter record. The data center segment contributed $1.833 billion in revenue, up 27% year-over-year and 11% quarter-over-quarter, accounting for 76% of total revenue.
Operating cash flow reached a record $639 million for the quarter, a year-over-year increase of 91.89%. However, under GAAP, net income attributable to shareholders was only $34.5 million, down 80.61% year-over-year, mainly due to acquisition-related expenses and non-cash items like stock-based compensation.
Prior to Huang’s remarks, MRVL’s closing market cap was around $192 billion. After the surge, it broke through $250 billion, and as of subsequent trading on June 5, it had climbed further to approximately $276.8 billion.
Key Calculation: To reach $1 trillion from about $250 billion, Marvell would need to grow its market cap roughly 3x; starting from $192 billion before Huang’s speech, it would require about a 4.2x increase.
The Connectivity Logic: Why Jensen Huang Is Bullish on Marvell
To assess the feasibility and requirements of a trillion-dollar valuation, it’s essential to unpack the core logic behind Huang’s prediction.
During his COMPUTEX 2026 keynote, Huang outlined a clear technical framework: AI computing is shifting from centralized, monolithic architectures to distributed, decoupled systems. Large language models and AI agents are no longer processed by a single chip or server; instead, their computational workloads are split across thousands of nodes in a data center, running in parallel.
"When you break a computation problem into many parts and distribute it across the data center, connectivity becomes indispensable," Huang stated. "That’s why Matt does so well, and why Marvell is so important."
From a supply chain perspective, Huang sees Marvell’s networking and connectivity chips as "essential" to AI infrastructure. This means Marvell’s market position is structurally resilient—regardless of which company supplies the compute chips, Marvell’s connectivity products serve as foundational infrastructure, especially as AI compute clusters scale.
This thesis is backed by data. Marvell’s data center segment continues to show robust growth: the FY2027 Q2 guidance midpoint projects revenue of about $2.7 billion, up roughly 35% year-over-year. The company has raised its FY2027 full-year revenue outlook to around $11.5 billion, a 40% increase, and set an FY2028 target of $16.5 billion, up $1.5 billion from the previous quarter’s guidance.
NVLink Fusion: The Technical Anchor for the Trillion-Dollar Trajectory
Huang’s optimism isn’t based solely on Marvell’s fundamentals; it’s rooted in the deepening technical partnership between NVIDIA and Marvell.
In March 2026, NVIDIA announced a $2 billion strategic investment in Marvell. The collaboration intensified through the NVLink Fusion platform: Marvell provides custom XPUs and vertical scaling network solutions compatible with NVLink Fusion, while NVIDIA contributes its Vera CPU, ConnectX network cards, and BlueField DPUs.
NVLink Fusion’s value proposition is clear: it offers cloud service providers a semi-customized path. Customers building AI data centers can strike a balance between NVIDIA GPUs and Marvell’s custom XPUs—not forced to buy NVIDIA’s full-stack solution, nor required to develop entirely independent alternatives. Huang candidly noted in his speech, "You don’t have to buy everything from us—just buy a part."
This framework lowers Marvell’s entry barriers in the AI chip market. Previously, Broadcom dominated custom XPU competition with a vast customer base and over 70% ASIC market share. NVLink Fusion gives Marvell differentiated access to large clients, allowing customers to avoid a binary choice between NVIDIA’s ecosystem and custom chip solutions.
Additionally, Huang’s technical perspective provides Marvell with incremental value regarding the transition timeline between copper and optical interconnects. He predicts, over the next 5–10 years, "Use copper wherever possible, and optical only where necessary." Marvell is one of the few companies offering comprehensive solutions for both copper and optical interconnects, giving it system-level resilience as AI data center connectivity needs continue to grow.
Custom Chip Business: Conditions for Achieving $10 Billion Revenue Guidance
Days before Huang’s trillion-dollar forecast, Marvell issued long-term guidance for its custom chip business in its FY2027 Q1 earnings report.
The company projects custom chip revenue will exceed $10 billion in FY2029. Morningstar analyst William Kerwin notes this guidance "implies a $5 billion incremental revenue contribution from this segment alone between FY2028 and FY2029."
From a TAM (Total Addressable Market) perspective, Marvell CEO Matt Murphy stated on the earnings call, "We have extensive custom partnerships with every major US hyperscale data center operator." Marvell’s market share in custom AI chips still lags Broadcom, but this gap also represents structural growth potential. If Marvell raises its share to about 20% before FY2029, the $10 billion revenue target is financially plausible.
However, achieving this goal depends on several conditions:
Condition 1: Diversification and growth of customer concentration. Currently, Marvell’s largest customer is AWS (Amazon Web Services), with the Trainium series AI ASICs as a key partnership. Heavy reliance on a single customer exposes Marvell to performance volatility if that customer’s capital expenditure fluctuates. Murphy has emphasized the need to diversify, aiming to secure orders from Microsoft Azure, Google Cloud, and Meta Platforms. If growth in FY2027 depends solely on AWS, the $10 billion target becomes much less attainable.
Condition 2: Support from TSMC’s advanced process and CoWoS capacity. Marvell’s custom chips are manufactured using TSMC’s 5nm and 3nm processes. Access to TSMC’s CoWoS advanced packaging capacity directly limits Marvell’s shipment capabilities. If demand for AI chips remains strong, CoWoS bottlenecks could become a constraint.
Condition 3: Sustainability of macroeconomic and capital expenditure cycles. US tech giants like Alphabet and Amazon are expected to invest over $700 billion in AI infrastructure in 2026 (up from about $400 billion in 2025). If AI capex growth slows between 2027 and 2029, custom chip revenue growth will be directly impacted.
Competitive Landscape: Custom XPU vs. General-Purpose GPU
To understand Marvell’s path to a trillion-dollar valuation, it’s necessary to evaluate the company within the broader AI chip competitive landscape.
In the custom XPU (ASIC) segment, Marvell and Broadcom are the two leading suppliers, but their business models differ sharply. Broadcom is known for its industry leadership and strong customer loyalty, with major clients like Google and Meta Platforms whose data center capex is rapidly expanding. Marvell’s current industry valuation (P/E ratio around 28, PEG ratio near 1) is at a discount compared to Broadcom, reflecting the market’s distinction between "challenger discount" and "leader premium."
Meanwhile, the custom XPU segment is undergoing a rebalancing with general-purpose GPUs. TrendForce forecasts ASIC shipments will grow 45% in 2026, far outpacing GPU growth at 16%, indicating broader market acceptance of custom designs. However, for training workloads, general-purpose GPUs retain irreplaceable advantages in performance and ecosystem maturity. Marvell’s ability to capture GPU market share with its custom XPU business depends on how sensitive customers are to cost and power consumption during inference.
Marvell’s NVLink Fusion ecosystem, built in partnership with NVIDIA, could provide a strategic differentiator versus Broadcom. Broadcom relies more on hyperscale customers developing their own solutions, while Marvell has achieved "certified partner" status within NVIDIA’s ecosystem. If the industry moves toward "heterogeneous computing," Marvell’s position could be more flexible than Broadcom’s. Conversely, if hyperscale clients like AWS increasingly push back against NVIDIA, Broadcom’s status as an independent supplier could become a competitive advantage.
External Acquisitions: Strategic Significance of Celestial AI and XConn Technologies
In February 2026, Marvell completed two major acquisitions, further strengthening its strategic position in AI data center connectivity.
Celestial AI (acquired for at least $3.25 billion) brought the breakthrough Photonic Fabric platform, designed for intra-system and rack-level optical connectivity. Marvell expects this technology will help capture the critical transition from electrical to optical interconnects, expanding its total addressable market in connectivity.
XConn Technologies (acquired for about $540 million) specializes in PCIe and CXL switch silicon. XConn has partnered with over 20 customers, with PCIe 5 and CXL 2.0 switches in mass production and PCIe 6 and CXL 3.1 in sampling. These products are expected to positively impact Marvell’s revenue in the second half of FY2027 and contribute about $100 million in incremental revenue in FY2028.
Risk Framework for the Growth Trajectory
Despite current market enthusiasm, it’s important to recognize factors that could constrain the trillion-dollar valuation.
Risk 1: Uncertainty in technical and industry direction. Huang’s thesis on "distributed/decoupled" AI computing is central to Marvell’s growth. If mainstream AI infrastructure shifts toward highly centralized, hyperscale clusters (reducing connectivity needs), Marvell’s valuation logic would require reevaluation.
Risk 2: Alignment between profitability and cash flow. While operating cash flow hit a record $639 million, GAAP net income was only $34.5 million, mainly due to non-cash acquisition amortization and stock-based compensation. The gross margin structure of custom chip business (similar to Broadcom, with high revenue growth often accompanied by margin pressure) needs ongoing scrutiny.
Risk 3: Structural vulnerability from customer concentration. Marvell’s custom chip revenue is heavily concentrated in AWS. If AWS adjusts its AI capex pace, Marvell’s quarterly performance and credibility of its billion-dollar revenue guidance could be directly affected.
Risk 4: Excessive sentiment premium in valuation. As of June 5, MRVL shares are up 272.36% year-to-date. Cantor Fitzgerald analyst CJ Muse commented after Huang’s remarks that the trillion-dollar valuation should be seen as "symbolic and aspirational," not as a precise financial forecast, and cautioned about technical overbought signals.
Conclusion
Marvell’s journey from its current $276.8 billion market cap to $1 trillion is not a predetermined path, but rather the result of a dynamic set of conditions being continuously met.
On the fundamentals side, $2.418 billion in quarterly revenue, $639 million in operating cash flow, billion-dollar custom chip guidance, and NVIDIA’s strategic investment form a solid foundation for growth. Strategically, Huang’s public endorsement at COMPUTEX 2026 positions Marvell as an "essential connectivity supplier" in AI infrastructure, with NVLink Fusion providing differentiated market positioning. Growth drivers include the acquisitions of Celestial AI and XConn, which expand Marvell’s addressable market in optical interconnects and PCIe/CXL switching.
However, significant constraints remain. Customer concentration (heavy reliance on AWS), competitive dynamics with Broadcom, sustainability of AI capex cycles, and potential sentiment-driven valuation premiums all introduce uncertainty. Additionally, whether NVIDIA will internalize more networking functions or build a parallel ecosystem through photonics investments is a variable to watch.
As the AI infrastructure industry shifts from scale expansion to efficiency optimization, Marvell’s ability to navigate technical divergence, evolving competition, and macro cycles—and ultimately reach a trillion-dollar market cap—hinges on one core variable: whether connectivity can truly evolve from an "auxiliary component" of AI compute to its "core bottleneck," thereby gaining structural pricing power.
