As of May 14, 2026, Marvell Technology closed at $182.58, representing an approximate 140% surge since March 5, starkly outperforming the S&P 500's 9% gain over the same period. The stock's near-vertical ascent began following the release of its fiscal Q4 2026 (ended January 31, 2026) earnings report on March 5. The company reported quarterly revenue of $2.219 billion, a 22% year-over-year increase, and non-GAAP EPS of $0.80, both exceeding market expectations.
The data center segment is the core growth engine. For the full fiscal year 2026, data center revenue surpassed $6 billion, growing 46% year-over-year. In the fourth quarter, data center revenue accounted for 74% of total revenue. The company has raised its financial outlook three times: the revenue forecast for fiscal 2027 has been revised upward from an initial $9.5 billion to nearly $11 billion (representing over 30% growth), and the fiscal 2028 revenue expectation now stands at approximately $15 billion.
Since the start of 2026, the stock's rally has accelerated. Year-to-date gains have exceeded 100%, with the stock closing at a record high 12 times. Particularly, on May 13 and 14, a confluence of catalysts propelled the stock to consecutive all-time highs, resulting in a cumulative gain of over 10% across the two trading sessions.
A Dense Convergence of Core Catalysts (1) AMD's Stake Disclosure Sparks M&A Speculation On May 12, AMD disclosed a stake in Marvell Technology in a regulatory filing, capturing intense market attention. On May 13, Marvell's stock surged 8.18% to close at $177.95, a record closing high. The market immediately speculated about potential strategic acquisition intent. While AMD has a strong presence in data center CPUs and GPUs, its AI interconnect technology is relatively weaker. Marvell is a leader in precisely that field. Although neither party has commented on a potential merger, AMD's stake itself signals to the market that this chip giant, with a market cap around $400 billion, is closely watching Marvell's strategic value.
(2) NVIDIA's $2 Billion Strategic Investment and NVLink Fusion On March 31, NVIDIA and Marvell announced a strategic cooperation agreement. NVIDIA is integrating Marvell into its AI factory and AI-RAN ecosystem via the NVLink Fusion platform and has made a $2 billion strategic investment. For Marvell, the significance of this deal extends far beyond the capital. First, the technical integration positions Marvell at the intersection of both GPU and ASIC ecosystems. NVLink Fusion allows customers to flexibly incorporate Marvell's custom chips and networking products on top of NVIDIA's architecture to build heterogeneous AI infrastructure. Second, the $2 billion investment represents NVIDIA's highest-level endorsement of Marvell's technological standing. RBC Capital analyst Pajjuri called it "a strong validation of Marvell's leadership in optical interconnect." Strategically, NVIDIA's move carries a deeper defensive logic—it has "completed a strategic encirclement in optical interconnect with $6 billion deployed within 30 days." As AI computing competition intensifies, NVIDIA is using capital ties to secure interconnect, a strategic infrastructure, ensuring its GPU empire is not isolated.
(3) Multiple Institutions Raise Price Targets, Highest at $210 On May 13, Marvell received a wave of bullish sentiment from Wall Street. BofA Securities significantly raised its price target from $125 to $200, maintaining a "Buy" rating and naming Marvell a "Top Pick." On the same day, B.Riley raised its target from $156 to $205; RBC Capital raised from $170 to $200. Although Goldman Sachs only raised its target from $100 to $125, it explicitly mentioned the potential upside from a possible collaboration with Google. Barclays had previously upgraded its rating to Overweight with a $150 target. B.Riley's latest target has even been raised further to $210.
BofA's analysis hinges on a significant upward revision of the AI networking total addressable market (TAM). The total AI connectivity TAM is projected to expand from about $6 billion in 2026 to $14 billion by 2030, with Ethernet transceivers contributing most of the incremental growth. The forecast for the AI data center systems market has been raised to approximately $1.7 trillion by 2030 (from a prior $1.4 trillion), with the AI accelerator segment alone expected to reach about $1.2 trillion.
(4) Google Custom Chip Collaboration & Amazon Trainium Expansion Simultaneously, developments with two major hyperscale cloud customers further strengthen the growth narrative for Marvell's custom chips. On April 20, multiple media reports indicated that Alphabet, Google's parent company, is in talks with Marvell to co-develop two new AI chips: a memory processing unit designed to work with Google's in-house TPUs, and a new TPU designed for AI inference tasks. Sources suggest the goal is to complete the design of the memory processing unit and begin pilot production before 2027. Google plans an initial production run of nearly 2 million such processors, complementing its annual TPU production of about 6 million units. Notably, Google is not looking to replace its existing partner Broadcom but rather to introduce a third party into its custom chip supply chain to diversify risk.
Following closely, Amazon announced an expanded partnership with Anthropic on April 20. Anthropic committed to investing over $100 billion in AWS technology over the next decade, with a key component being the Trainium chip co-designed by Amazon and Marvell. According to Wells Fargo data, Anthropic has already deployed over 1 million Trainium chips on AWS. As Trainium 3 ramps up in the second half of 2026, the corresponding computing capacity is expected to reach 2GW. J.P. Morgan analysts noted in a report: "Marvell is Amazon's long-term custom chip design partner for the Trainium project, currently accelerating the volume production of Trainium 3 and is poised to further expand its share in the future Trainium 4."
(5) Polariton Acquisition Strengthens Silicon Photonics Position On April 22, Marvell announced the acquisition of Swiss high-speed optical chip company Polariton Technologies, a pioneer in the commercialization of plasmonic electro-optic technology. Polariton's plasmonic modulators boast leading specifications: electro-optic bandwidth exceeding 110GHz, demonstrated single-channel rates surpassing 400Gbit/s, with a target to control 400G channel power consumption below 0.5pJ/bit. Strategically, this acquisition targets the technological leap from "1.6T to 3.2T." As AI workloads drive exponential bandwidth demand growth, data center scaling and data center interconnect scenarios urgently require the commercialization of 3.2T and higher bandwidth standards. Polariton's plasmonic modulation technology is a key enabler for this transition. Combined with the photonic interconnect technology gained from the prior acquisition of Celestial AI for up to $5.5 billion, Marvell is building a comprehensive optical interconnect portfolio covering "electro-optic conversion, photonic integration, and switch chips."
Dual Engines: Interconnect & Custom Chips In BofA analyst Vivek Arya's bullish framework, Marvell's high-value growth stems from two core dimensions: the dual expansion of the Ethernet transceiver market and the custom chip market. Regarding network interconnect, BofA expects the Ethernet transceiver market to add $10 billion by 2028, directly driving the DSP market size up by approximately $2 billion. Given Marvell maintains a roughly 60% to 70% share in the 800G and 1.6T transceiver DSP market, this alone could bring over $1.2 billion in additional revenue. Consequently, BofA raised its total sales estimates for fiscal 2028 and 2029 to $15.17 billion and $20.02 billion, respectively, and its non-GAAP EPS estimates (excluding stock-based compensation) to $5.60 and $7.80, respectively.
On the custom chip front, Marvell has secured design wins for 18 cloud provider projects, including Amazon's Trainium series, Microsoft's Maia AI accelerator, and Meta's data processing unit, with annualized custom chip revenue around $1.5 billion. Marvell CEO Matt Murphy has explicitly stated the strategic goal of "achieving a 20% share in the custom AI chip market." Goldman Sachs analyst James Schneider's team noted that a potential partnership with Google could further boost custom ASIC revenue, making it a key focus for the upcoming earnings call.
From a valuation perspective, BofA's $200 target is based on approximately 30 times the estimated fiscal 2028 EPS (including stock-based compensation). Although Marvell's current TTM P/E is around 59x, placing it alongside Broadcom among the "most expensive semiconductor stocks," as fiscal 2028 revenue expectations converge toward $15 billion and EPS toward the $5-$7.80 range, the forward valuation is expected to be digested rapidly.
At the industry macro level, global AI data center construction demand has remained robust since 2026, with tech giants worldwide raising their full-year capital expenditure guidance. In Q1 2026, the combined capex of the four major cloud service providers (Google, Microsoft, etc.) reached $128.8 billion, an 81% year-over-year increase. IDC predicts global data center semiconductor revenue will reach $477.1 billion in 2026, contributing $843.2 billion by 2030, accounting for nearly half of the entire semiconductor market. Data from J.P. Morgan Asset Management shows AI-related trades now constitute 57.3% of the S&P 500's weight, with the semiconductor sector leading at 18.2%. As a core supplier in AI interconnect and custom chips, Marvell is positioned at the center of this historic wave.
Upcoming Key Catalyst: Q1 2027 Earnings Preview Marvell is scheduled to report its fiscal Q1 2027 (ended April 30, 2026) earnings on May 27. Wall Street expects quarterly EPS of approximately $0.79, a 27.4% year-over-year increase, and revenue of about $2.4 billion, also up around 27%. This aligns with the company's prior guidance of "$2.4 billion, plus or minus 5%." However, market expectations for this report extend beyond merely meeting estimates. Goldman Sachs analysts explicitly stated that the most noteworthy items on the earnings call will include: updates on the potential partnership with Google, revised financial guidance for fiscal 2027-2028, and the impact of short-term gross margin fluctuations. RBC Capital analyst Pajjuri pointed out that while the custom XPU (accelerated processor) trend is strong, tight wafer supply may constrain near-term upside. With the current stock price already reflecting significant optimism, the "substance" of the May 27th earnings will directly determine whether Marvell can sustain its remarkable rally.
Potential Risks On the flip side of the optimistic narrative, Marvell investors must cautiously consider the following risks: Valuation Risk: The current TTM P/E is approximately 59x. Even based on the most optimistic fiscal 2028 EPS estimate of $7.80, the forward P/E remains above 23x. Should AI capital expenditure growth slow or company execution falter, the high valuation poses a significant risk to the stock price. Customer Concentration & Competition Risk: Amazon is Marvell's largest custom chip customer. Although a five-year cooperation agreement exists, strategic shifts by a single customer could have a material impact. Concurrently, Broadcom is expanding aggressively with a target of over $100 billion in annual AI chip revenue, and MediaTek has entered Google's TPU supply chain. Furthermore, the custom ASIC path itself faces strong competition from NVIDIA's next-generation GPU platforms. If GPU cost-performance continues to improve, some cloud customers might scale back ASIC deployments. Execution & Supply Chain Risk: Marvell's rapid succession of acquisitions, including Celestial AI (up to $5.5 billion) and Polariton, presents integration challenges. Simultaneously, sustained tightness in advanced wafer capacity could constrain the ramp-up speed of the custom chip business. Negotiations with Google have "not yet resulted in a formal contract," introducing uncertainty regarding finalization.
Conclusion Marvell has undergone a profound transformation from a fabless semiconductor company to a dual leader in AI data center interconnect and custom chips. The "three pillars" of NVIDIA's $2 billion strategic investment, the acceleration of the Amazon Trainium ecosystem, and the potential Google partnership form the foundation for the company's growth over the coming years. The silicon photonics technology acquired through Celestial AI and Polariton further establishes long-term competitive barriers for the evolution toward 3.2T and higher bandwidths.
However, Marvell, now a company with a market cap in the hundreds of billions, is no longer an overlooked "hidden champion." It has become a highly focused AI core holding for global investors. The market views it as a "neutral arms dealer" capable of serving both the GPU and ASIC AI computing ecosystems. The Q1 earnings report on May 27 will be a crucial test of these lofty expectations. With multiple catalysts already reflected in the stock price, can the company reignite the upward engine with better-than-expected results and guidance? At the current, intensely contested price level, the answer is imminent.
Comments