What happened
Reuters reported that Marvell Technology shares fell more than 8% in early trading on August 28 after the chip designer released results and forecasts that investors considered insufficiently strong relative to expectations. The central concern was the timing of revenue from a newly announced Google custom-chip deal, which Marvell says could generate up to $120 billion through fiscal 2033.
Reuters, in a report republished by The Economic Times, said Marvell shares fell more than 8% to $221.60 in early trading after the company reported results that were described as solid but failed to provide the fresh evidence investors wanted about long-term growth. The decline put Marvell on course to lose more than $17.4 billion in market value if the losses held, according to the report. This was a market reaction to expectations around an AI-related business opportunity, rather than a report of a failure, cancellation, or technical problem involving Marvell’s chips or Google’s systems.
Reuters said Marvell has benefited from the boom in spending on AI infrastructure as large technology companies adopt custom chips to seek better cost efficiency and performance. The company’s shares had nearly tripled during the year before the reported selloff. Reuters also said that total Big Tech AI spending was expected to exceed $740 billion in the year covered by the report. The source did not independently establish that spending figure, nor did it provide a detailed breakdown of how much of it would flow to Marvell.
The immediate focus was a newly announced Google custom-chip deal that Marvell said could generate up to $120 billion in revenue through fiscal 2033. Reuters reported that investors wanted to know how quickly the agreement would begin contributing meaningfully to revenue. Marvell chief executive Matt Murphy said the company’s custom-revenue targets through fiscal 2028 already reflected some Google-related revenue, while the contribution would become much more significant in fiscal 2029. The report did not provide the deal’s full commercial terms, delivery schedule, chip volumes, pricing, margins, or a public commitment from Google confirming the projected revenue.
Reuters reported that Marvell expects revenue to grow by about 45% in fiscal 2027 and reach about $18 billion in fiscal 2028, helped by greater data-center revenue. Those forecasts were not enough to satisfy all market expectations, according to the report. At least eight brokerages raised their price targets after the results, with a median target of $275, based on LSEG data cited by Reuters. These targets are analyst estimates, not independently verified outcomes, and they do not establish that Marvell will achieve the projected growth or that the Google agreement will reach its maximum value.
Source details: m.economictimes.com ↗
Why it matters
The report illustrates how heavily semiconductor valuations now depend on the expected payoff from AI infrastructure spending, not only on current earnings. It also highlights the gap between the headline size of a major custom-chip agreement and the slower, staged revenue contribution investors may see in the near term.
The report matters because it shows that the AI infrastructure boom is being evaluated through a demanding financial lens. A large projected contract can support a company’s long-term narrative, but investors may still penalize the company if the expected revenue arrives later than anticipated. In Marvell’s case, Reuters described a mismatch between the size of the Google opportunity and the relatively limited near-term contribution reflected in management’s guidance.
The reported deal also illustrates the role of custom chips in the AI supply chain. Reuters said major technology companies are adopting custom silicon to pursue cost efficiency and performance. That creates opportunities for chip designers such as Marvell, but it also means that future revenue may depend on customer deployment schedules, product qualification, manufacturing execution, and the pace at which AI data-center capacity is built. The source did not independently confirm any of those operational milestones.
For the public and for businesses that depend on AI services, the financial issue has practical implications. Custom-chip programs can influence the cost, availability, and scale of AI computing, but the report offers no evidence that the Google agreement has already changed consumer access, service prices, model capability, or data-center performance. Its significance is primarily industrial and financial: it provides a window into how the infrastructure companies supporting AI are being valued and how quickly investors expect that demand to translate into reported earnings.
Reuters also reported that Marvell trades at a premium to rival Broadcom, with a 12-month forward price-to-earnings ratio of 58.41 compared with 32.15, using LSEG data. That comparison helps explain the sensitivity of the stock to any perceived gap between expectations and guidance. It does not by itself show that Marvell is overvalued, that Broadcom is a direct substitute for every Marvell product, or that either company’s AI-related outlook will prove accurate. Those points remain unresolved from the supplied source.
What to watch next
The key unresolved issue is when Google-related revenue will become material and how much of Marvell’s existing forecasts already include it. Reuters also identifies Marvell’s exposure to Microsoft, AI connectivity demand, future earnings expectations, and its valuation relative to Broadcom as factors investors may continue to reassess.
The most important next signal is the timing and size of actual Google-related revenue. Reuters reported management’s view that some contribution is included in targets through fiscal 2028 and that the effect will be much larger in fiscal 2029. Investors will likely look for future disclosures that separate Google revenue from Marvell’s broader custom-chip business, but the supplied report does not say whether the company will provide that level of detail or when it will do so.
A second issue is whether Marvell’s broader AI exposure develops as described. Reuters cited prospects involving Microsoft and AI connectivity as part of the company’s longer-term case. The report did not identify specific Microsoft products, contracts, deployment dates, or revenue figures, so those prospects should be treated as reported possibilities rather than confirmed business outcomes. It also did not establish whether connectivity demand will grow at the same rate as demand for AI accelerators or custom chips.
Investors will also watch whether the company’s fiscal 2027 and fiscal 2028 forecasts are met. Reuters reported expected revenue growth of about 45% in fiscal 2027 and approximately $18 billion in fiscal 2028, but the report did not provide enough underlying information to assess the assumptions behind those figures. The source also cited a Melius Research view that Marvell could eventually reach $20 in earnings per share before the end of the decade; that is an analyst projection, not a confirmed result.
Finally, the headline value of up to $120 billion should not be confused with booked or realized revenue. The report does not independently confirm the maximum value, the customer’s obligation to purchase at that level, the timing of recognition, or the profitability of the work. Until those unknowns are clarified through company disclosures or other reliable reporting, the Google agreement is best understood as a large long-term opportunity whose financial payoff remains uncertain.