Marvell Technology, Inc. is a market -based capitalization company of $ 60 billion which designs, develops and markets a wide range of circuits and integrated solutions for the construction of data infrastructure. Essentially, its products are used on various final markets which are strongly focused on data centers. Despite the publication of income and non-GAAP 4Q EPS figures which beat street estimates and with 1Q guidelines for the 201026 financial year, a little above consensus, the stock dropped by 20%. Normally, such results would trigger a positive reaction, but the sale has coincided with a weak feeling motivated by the uncertainty of American trade policy and the recent disappointing income from Nvidia. As a result, Marvell’s YTD losses almost reached 38%. In my opinion, this heavy slowdown is a reaction on the exaggerated market.
Marvell technology: player in the undervalued data center
Although the stock is not yet at deep negotiation levels, its multiple evaluation have compressed on a term basis. Even if Marvell appears at a richly price compared to p / e before standards, I think that the prospects for the growth of the company remain strong, especially since its multiple PEG is now less than 1x for the next two years. This suggests that Marvell does not deserve such a pessimistic reaction. I expect the trend in innovation in secular AI to stimulate the demand for higher data center, in particular, while the company accelerates volume shipments for XPU to its hyperscalers. The stock will probably rebound in the coming months. Let me explain my justification more in this article.
Marvell’s growth momentum remains robust. Total turnover increased by 27.4% in an annual shift in 4q for the 201025 financial year, widely fired by a stellar increase of 78.5% in annual slip of the data center. With management guidelines for 1q FY2026 predicting an increase of 61.5% in annual slip of the data center income, I do not expect a net slowdown from here. It is clear that the main engine of Marvell’s first line is the turnover of the data center, which now represents 75% of the total returned mixture which, I think, is a strong indicator of future trends. In addition, more than half of this income from the data center are supplied by AI. The company has generated $ 2.5 billion in IA -related income during the 20125 fiscal year, considerably defeating its $ 1.5 billion guidelines compared to the April 2024 AI day, and management expects that IA 2026 revenues will be much 2.5 billion dollars, with street consensus at 3 billion dollars. This level of more than $ 2.5 billion suggests that the data center market will probably continue to grow strongly in EFY26, even if there is a potential drop in AI calculation demand.
The expected growth is mainly motivated by reducing volume shipments for XPUs through hyperscalers. Despite short -term uncertainty, such as the news according to which Microsoft has canceled certain data center leases, I see it as a simple noise; Hyperscalers will probably continue to invest massively in calculation capacity in the coming years. Marvell has experienced accelerated growth in its personalized ASIC activities in a similar way to the dynamics of the Broadcom market, largely due to partnerships with major hyperscalers. A key partnership is with Meta Platforms, where Marvell signed a 4Q FY2025 agreement to provide ASIC, a firmware and personalized software. This agreement could be a major income engine because META is expanding its data center footprint, management at META suggesting a capital expenditure of 60 to 65 billion dollars for the EFY2026 while developing Llama 4 and increases their AI platform. In addition, Marvell has already pointed out volume production from its personalized AI silicon programs with established customers like Amazon and Alphabet, which should be high -level FY2026 and FY2027. The long-term advantage of these partnerships is the potential for multi-generational versions, which offers a certain protection against recession, because these hyperscal capital expenses, in particular in AI, are probably not reversed given the intense competition and their robust expenditure capacities.
I believe it is clear that if Nvidia maintains a dominant position in the market sector due to the flexibility of its chipsa highlighted by the CEO Jensen Huangthe Real Growth Driver for Marvell is the need for hyperscalers of personalized solutions. Even players in Chinese AI space recognize efficiency gains motivated by innovations like those of Deepseek. This continuous desire for algorithmic improvements according to which lower costs should benefit from the calculation of personalized AI, helping Marvell to maintain and even increase her market share. Although Broadcom is a great rival, I consider Marvell’s profitable commercial model and price flexibility as key advantages in competition on a more attractive price / value with hyperscalers. Marvell has already obtained design victories for personalized nicks and personalized CXL memory solutions with major hyperscalers such as Meta platforms, preparing the track for remarkable operational tail winds over the next 1 to 2 years as the momentum is built in the personalized silicon space.
The emphasis on data centers has temporarily resulted in the gross margin adjusted for the sale of lower margin for the first quarter of the 4025 financial year, a slight decrease compared to the 63.9% of the fourth quarter 201024, largely due to personalized silicon volumes raised in the data centers segment. However, I consider a margin of 60.1% as acceptable because it is the target beach of Marvell, even if it leaves a limited safety margin. The company continues to maintain a perspective of resilient margin; Although its non -Gaap gross margin has normalized without a significant decrease, its margin of operating income not in accordance with PCAP extended in particular to the fourth quarter FY2025, and the prospects of 1q for the 2010 exercise suggest that the margins will remain relatively stable compared to the previous quarter. This indicates that Marvell benefits from a strong operating lever effect, which helps compensate for lower margin entries as the company of the ladder ladder data center. Management has reiterated a long-term objective for non-gap operating income margins of 38% to 40% (compared to 33.7% in the last quarter), which is a key anchor for me because it shows the optimism of the company as to the scaling of its personalized AI calculation and its high-speed networking activities. I expect the margins to continue to improve in the EFY2026 while the production volume increases for the XPU products of its hyperscalers.
From a fundamental perspective, I think that actions are considerably undervalued, in particular when they consider their historical negotiation premiums. The action was negotiated at a 40x P / E ratio before the crisis (when the equity price was greater than $ 60 a few weeks ago). Now, the image is different because Marvell is negotiated at only 25 times the profits of next year, and what is more important, the PEG ratios for the next few years are much less than 1x. During the comparison of the P / E ratio at 3 years from 14.5x with its direct chip peers, only Qualcomm is negotiated less at 12.9x, while Broadcom, analog devices and the NVIDIA negotiate considerably higher at 21.7x, 19.3x and 18.6x respectively, while the AMD and Intels are negotiated relatively in line with Marvell.
Marvell technology: player in the undervalued data center
(Spread Alpha)
I think this multiple should come back realistically at least around 17x; First, because it exchanged considerably above that of the last quarters, and secondly, because it is directly comparable to Broadcom, which specializes in personalized ASIC and should see similar growth, if not more, because Marvell is at a lower base because of its smallest size. I believe that this evaluation gap between Broadcom and Marvell should tighten. Given my 17x multiple hypothesis, we would consider a target price of around $ 81 with a safety margin of around 17%, which, I think, is much given the hypothesis of multiple conservative expansion.
I also adopted a complete DCF approach to train my expectations for the company. In this DCF model for Marvell, I assumed a growth period of growth over 5 years, with a discount rate of 10% and a terminal growth rate of 5%.
Marvell technology: player in the undervalued data center
(Spread Alpha)
Regarding the front line line, I expect it to go from approximately $ 8.15 billion in $ 1 to $ 14 billion by the 5th year, mainly driven by the activity of the company data center. Given the capital expenses of hyperscalers such as Meta, Amazon and Alphabet specially for ASICS, NAVS and CXL memory solutions personalized expect that the Marvell income mixture remains tilted to the products of the data center, supplying continuous superior dynamics. On the margin front, the model assumes that the net margins will improve from the 1920s in the mid -1930s by the end of the forecast period, reflecting a better operating lever effect as the volumes of the data center increases, and this is aligned with the objective of longer -term operating income margin of the management of 3840%. Income growth trajectory and expansion of margins are largely in line with Wall Street expectations. Finally, I expect the CAPEX Net to go from around $ 328 million to $ 946 million by the 5th year while the company is investing in the manufacture and design of advanced nodes for the calculation of the new generation AI. However, Marvell’s scale and partnerships should allow him to absorb these costs without compromising the expansion of margins.
Based on these hypotheses, the DCF model gives a target price of $ 87, which implies 22% of the increase in current levels, which, I believe, still improves my upward thesis.
In conclusion, I believe that the decrease of almost 30% of Marvell is exaggerated, in particular taking into account his 4q solid income and his impressive prospects. Recent sale seems to be more motivated by a wide change in market feeling than by any real weakness of the company. The multiple evaluation have considerably compressed, creating an attractive purchasing opportunity for a company that is still in a high growth phase. I do not see any clear sign that demand in its final markets is blocking, Marvell has firmly positioned himself as a pure game in the space of the data center. Since this segment is about to develop more in the coming years, I expect the multiple Marvells to go back to or even exceed the past levels. For these reasons, I remain optimistic about Marvell at her current levels.
This article appeared for the first time on Gurufocus.
