A two-month surge of 140%, breaking the record 12 times this year! Bank of America Goldman Sachs Group, Inc. has raised its target price again. Is the bullish outlook for Marvell Technology, Inc. (MRVL.US) strengthening?

As of May 14, 2026, Marvell closed at $182.58, representing a significant increase of about 140% since March 5, while the S&P 500 index only rose 9% during the same period. Since the announcement of the financial report for the fourth quarter of the 2026 fiscal year (ending January 31, 2026) on March 5, Marvell's stock price has embarked on a nearly straight upward trend. The revenue for the quarter was $2.219 billion, a year-on-year increase of 22%, and the non-GAAP earnings per share were $0.80, both exceeding market expectations. The data center is the core growth engine. The full-year data center revenue for the 2026 fiscal year exceeded $6 billion, an increase of 46% year-on-year, and in the fourth quarter, data center revenue accounted for 74% of total revenue. The company has raised its financial outlook for the third time: revenue for the 2027 fiscal year is now expected to increase from the initial estimate of about $9.5 billion to nearly $11 billion (a year-on-year increase of over 30%), and revenue for the 2028 fiscal year is expected to reach around $15 billion. Since the beginning of 2026, the stock price has been accelerating. The cumulative increase for the year has surpassed 100%, with 12 record high closings. Especially on May 13th and 14th, multiple catalysts resonated, leading to consecutive historic highs in the stock price, with a cumulative increase of over 10% over the two trading days. Key catalysts resonate intensively 1. AMD discloses holdings, triggering "merger speculation" On May 12, AMD disclosed its holdings in Marvell in a regulatory filing, attracting strong market attention. On May 13, Marvell's stock price surged by 8.18%, closing at $177.95, setting a new all-time high. The market immediately speculated about a potential strategic acquisition - as AMD has a strong presence in the data center CPU and GPU fields but lacks in AI interconnection technology, Marvell happens to be a leading player in this area. Although both parties have not commented on any merger intentions so far, AMD's holdings alone signify to the market the strategic value that this $400 billion chip giant is closely monitoring in Marvell. 2. NVIDIA's $1 billion strategic investment and NVLink Fusion On March 31, NVIDIA and Marvell announced a strategic partnership agreement, where NVIDIA will incorporate Marvell into its AI factory and AI-RAN ecosystem through the NVLink Fusion platform and make a $2 billion strategic investment. For Marvell, the significance of this deal goes beyond the financial investment. Firstly, the technical integration places Marvell at the intersection of the two major ecosystems of GPU and ASIC. NVLink Fusion allows customers to flexibly introduce Marvell's custom chips and network products on top of the NVIDIA architecture to build a heterogeneous AI infrastructure. Secondly, the $2 billion strategic investment is NVIDIA's highest level of recognition of Marvell's technological position. RBC Capital analyst Pajjuri commented, calling it a "strong validation of Marvell's leading position in the optical interconnect field". From a strategic perspective, NVIDIA's strategy carries a deeper defense logic - its "spending $6 billion within 30 days to complete its optical interconnect strategic encirclement". With the AI computing competition intensifying, NVIDIA is using capital ties to lock in interconnection, a strategic infrastructure, to ensure its GPU empire is not isolated. 3. Multiple institutions raise target prices, with the highest target price at $210 On May 13, Marvell saw a surge of bullish sentiments from Wall Street. Bank of America Securities raised its target price from $125 to $200, maintaining a "buy" rating and listing Marvell as a "top pick" stock. On the same day, B. Riley raised its target price from $156 to $205; RBC Capital raised its target from $170 to $200; and although Goldman Sachs only raised its target price from $100 to $125, it explicitly mentioned the potential upside that a Google collaboration could bring. Barclays had previously upgraded its rating to overweight with a target price of $150. B. Riley's latest target price has further increased to $210. Bank of America's core rationale lies in the significant increase in the addressable market size for AI networks: the total addressable market for AI connectivity is expected to expand from about $6 billion in 2026 to $14 billion in 2030, with most of the incremental contribution coming from Ethernet transceivers; the forecast for the AI data center systems market has been raised to about $1.7 trillion in 2030 (up from $1.4 trillion), with the AI accelerator sub-market expected to reach about $1.2 trillion alone. 4. Google's custom chip collaboration and Amazon's Trainium expansion Meanwhile, the positioning of Marvell's custom chip growth logic was further strengthened by the enhanced layouts of two large-scale cloud customers. On April 20, according to multiple media reports, Alphabet, Google's parent company, has been in talks with Marvell to jointly develop two new AI chips - one for a memory processing unit that is intended to work in conjunction with Google's TPU, and another for a new type of TPU specifically designed for AI inference tasks. Insiders revealed that the two companies aim to complete the design and start production of the memory processing unit by 2027, with Google planning to initially produce nearly 2 million of these processors, complementing its yearly output of about 6 million TPUs. It is worth noting that Google is not looking to replace its current partner, Broadcom, but rather introduce a third-party force into its custom chip supply chain to diversify risks. Shortly after, on April 20, Amazon announced an expansion of its partnership with Anthropic. Anthropic has committed to invest over $100 billion in AWS technology over the next ten years, and a key component of this agreement is Amazon's collaboration with Marvell to design the Trainium chip. According to data from Fidelity Bank, Anthropic has deployed over a million Trainium chips on AWS, and with the mass production of Trainium 3 in the latter half of 2026, the corresponding computing capacity is expected to reach 2GW. Analysts at J.P. Morgan noted in an investment report, "Marvell is Amazon's long-term custom chip designer partner in the Trainium project, currently accelerating the mass production of Trainium 3, and is expected to further expand market share in future Trainium 4." 5. Acquisition of Polariton strengthens silicon photonics layout On April 22, Marvell announced the acquisition of Polariton Technologies, a Swiss high-speed optical chip company and a pioneer in commercializing plasma optoelectronic technology worldwide. Polariton's plasma modulators boast several leading indicators: an electro-optical bandwidth exceeding 110GHz, a measured single-channel speed surpassing 400Gbit/s, with a target channel power consumption kept under 0.5pJ/bit. Strategically, this acquisition is aimed at the "1.6T to 3.2T" technology intergenerational leap - as AI workloads drive exponential growth in bandwidth demand, the horizontal expansion and interconnection scenes of data centers urgently require the commercialization of 3.2T and higher bandwidth standards, and Polariton's plasma modulation technology is the key to unlocking this door. Coupled with the acquisition of Celestial AI, which obtained photon interconnect technology worth as much as $5.5 billion, Marvell is currently constructing a complete chain covering "electro-optical conversion, photon integration, and exchange chips" in the optical interconnection blueprint. Interconnect and custom chip dual-driven In the optimistic framework of Bank of America analyst Vivek Arya, Marvell's high-value growth is derived from two core dimensions: the expansion of the Ethernet transceiver market and the custom chip market. Regarding network interconnectivity, Bank of America expects the Ethernet transceiver market to increase by $10 billion by 2028, directly driving an approximate $2 billion increase in DSP market size. Considering Marvell's 60-70% market share in the 800G and 1.6T transceiver DSP market, this alone is expected to add over $1.2 billion in additional revenue. Bank of America has thus raised its sales expectations for the fiscal years 2028 and 2029 to $15.17 billion and $20.02 billion, respectively, with non-GAAP earnings per share expectations (excluding equity compensation) raised to $5.60 and $7.80, respectively. In terms of custom chips, Marvell has secured design orders from a total of 18 cloud providers, including Amazon's Trainium series, Microsoft's Maia AI accelerator, and Meta's data processing units, with an annual revenue of about $1.5 billion from custom chip business. Marvell CEO Matt Murphy has set a strategic goal of capturing a 20% share of the custom AI chip market. Analysts at Goldman Sachs point out that potential cooperation with Google is expected to further boost revenue from custom ASIC chips, becoming a key focus of the earnings call. From a valuation perspective, Bank of America's $200 target price is based on an estimated EPS of about 30 times in 2028 (including equity compensation). Despite Marvell's current TTM P/E ratio of about 59 times, ranking alongside Broadcom as one of the "most expensive semiconductor stocks," as revenue expectations for fiscal year 2028 are reaching $15 billion and the EPS is converging around the $5-7.8 range, long-term valuations are expected to be quickly digested. At the industry macro level, global demand for AI data center construction has remained consistently strong since 2026, with major technology giants both domestically and internationally raising their full-year capital expenditure guidance. In the first quarter of 2026, the combined capital expenditures of Google, Microsoft, and the other three major cloud service providers amounted to $128.8 billion, an 81% year-on-year increase. IDC predicts that global revenue from data center semiconductors in 2026 will reach $477.1 billion, and by 2030, it is expected to contribute $843.2 billion to the entire semiconductor market, almost dominating half of the market. Data from J.P. Morgan Asset Management shows that AI transactions already account for 57.3% of the S&P 500 index, with the semiconductor sector leading with an 18.2% share. As a core supplier of AI interconnectivity and custom chips, Marvell is positioned at the center of this historic wave. Upcoming key points: Outlook for the May 27th financial report Marvell has confirmed it will release the financial report for the first quarter of the 2027 fiscal year (ending April 30, 2026) on May 27. Wall Street expects earnings per share for the quarter to be around $0.79, an increase of approximately 27.4% year-on-year, and revenue to be around $2.4 billion, an increase of about 27%. This expectation aligns closely with Marvell's previous guidance of "around $2.4 billion plus or minus 5%". However, market expectations for this financial report go far beyond just meeting the standard. Goldman Sachs analysts explicitly point out that the most noteworthy items in the earnings call will include: the latest developments in potential collaboration with Google, updated performance guidance for the 2027-2028 fiscal years, and the impact of short-term fluctuations in gross profit margins. RBC Capital analysts also highlight the strong trend in custom XPUs (accelerators), but note that tight wafer supplies may restrict short-term upside potential. With the stock price already heavily inflated with optimistic expectations, the "golden content" of the May 27 financial report will directly determine whether Marvell can continue its legendary uptrend. Potential risks On the other side of the optimistic narrative, Marvell investors must cautiously face the following risks: High valuation risks: The current TTM P/E ratio is around 59 times. Even with the most optimistic EPS of $7.80 for the fiscal year 2028, the forward P/E ratio is still over 23 times. If the growth rate of AI capital expenditures slows down or the company's execution deviates, the high valuation could be a sharp blade hanging over the stock price. Customer concentration and competition risks: Amazon is Marvell's largest custom chip customer, and although they have a five-year partnership agreement, strategic adjustments by a single customer could have a significant impact. At the same time, Broadcom is aggressively expanding its AI chip revenue target to over $100 billion annually, while MediaTek has entered Google's TPU supply chain. In addition, the custom ASIC route itself faces strong competition from NVIDIA's next-generation GPU platform - if the cost-effectiveness of GPUs continues to rise, some cloud customers may reduce the deployment scale of ASICs. Execution and supply chain risks: Marvell has made several consecutive acquisitions, including Celestial AI (up to $5.5 billion) and Polariton, and the risks associated with integration cannot be ignored. Additionally, high-end wafer capacity remains tight, potentially limiting the speed of scaling up in the custom chip business. The negotiations with Google have so far not resulted in a formal contract, introducing uncertainties about whether the deal will ultimately materialize. Conclusion From being a semiconductor company without wafer fabrication facilities to a dual-leader in AI data center interconnectivity and custom chips, Marvell has undergone a profound transformation in its corporate positioning. The "three horsemen" of NVIDIA's $2 billion strategic investment, Amazon's Trainium ecosystem acceleration, and potential collaboration with Google form the cornerstone of the company's growth in the coming years. The acquisitions of Celestial AI and Polariton, which focus on silicon photonics technology, further establish long-term competitive barriers for Marvell's evolution towards 3.2T and higher bandwidths. However, the $100 billion market cap of Marvell is no longer an ignored "hidden champion" but a core target of global investors in AI. The market now sees it as a "neutral arms dealer" that can serve both the GPU and ASIC AI computing ecosystems. The first-quarter financial report on May 27 will be a key test of these high expectations - after multiple catalysts have been fully reflected in the stock price, can the company ignite its uptrend again with better than expected performance and guidance? At the current price range where both bullish and bearish sentiments are intensifying, the answer will soon be revealed.