--- title: "NVIDIA bets $2 billion, Marvell Technology's 'AI Interconnect' story may have just begun." type: "Topics" locale: "zh-CN" url: "https://longbridge.com/zh-CN/topics/39670331.md" description: "If this transaction is viewed merely as a financial investment, it would be an underestimation. NVIDIA has invested $2 billion in Marvell Technology. Following the news, Marvell's stock price rose by approximately 7% at one point, and NVIDIA's stock price also strengthened in tandem. More importantly, this money isn't simply buying a gesture; it directly targets the communication, interconnection, and photonics upgrades within AI data centers. In other words, NVIDIA is adding a piece to its AI ecosystem that is most easily overlooked but increasingly crucial..." datetime: "2026-04-01T12:15:04.000Z" locales: - [en](https://longbridge.com/en/topics/39670331.md) - [zh-CN](https://longbridge.com/zh-CN/topics/39670331.md) - [zh-HK](https://longbridge.com/zh-HK/topics/39670331.md) author: "[Elena mi](https://longbridge.com/zh-CN/profiles/25148023.md)" --- > 支持的语言: [English](https://longbridge.com/en/topics/39670331.md) | [繁體中文](https://longbridge.com/zh-HK/topics/39670331.md) # NVIDIA bets $2 billion, Marvell Technology's 'AI Interconnect' story may have just begun. If this transaction is viewed merely as a financial investment, it would be an understatement. NVIDIA has invested **$2 billion** in Marvell Technology. Following the news, Marvell's stock price rose by approximately **7%** at one point, and NVIDIA's stock price also strengthened in tandem. More importantly, this money isn't simply buying a gesture; it directly targets **communication, interconnectivity, and photonics upgrades within AI data centers**. In other words, NVIDIA is adding a piece to its AI ecosystem that is most easily overlooked but increasingly crucial. I'm more inclined to interpret this deal as: **NVIDIA is making early-stage arrangements for the AI infrastructure of the post-training, post-single-card era.** In the past, everyone focused on GPU computing power. Now the problem has become: after the computing power is delivered, how do chips connect, how is memory managed, and how does the network run faster and more power-efficiently? This collaboration will further integrate Marvell's semi-custom chips and optical interconnect technology with NVIDIA's central processors and networking equipment, focusing on solving bandwidth and energy efficiency bottlenecks, particularly around the compatibility of **NVLink Fusion**. However, the reason this company was chosen by NVIDIA isn't because it suddenly caught a trend, but because it has been actively transforming itself over the past two years. Marvell acquired the networking equipment company **XConn Technologies** in early 2026 and the photonics startup **Celestial AI** in late 2025, with a very clear goal: to push itself from a traditional semiconductor supplier onto the entire chain of **AI interconnect, CPO, SiPho, and CXL memory controllers**. These moves are part of Marvell's efforts to strengthen its AI data center infrastructure layout. This step is actually quite smart. Because the most dangerous aspect for Marvell before wasn't a lack of technology, but **over-reliance on a few major customers, especially the custom chip business for cloud providers**. Marvell has already shifted its focus from pure XPU/ASIC orders towards broader AI interconnect solutions. Now, coupled with NVIDIA's investment, it's essentially pushing the company from a point-solution business to a platform-type supplier. This is also why the market sees it as a company with the potential for re-rating, rather than just a cyclical stock. Looking at performance, Marvell's transformation isn't just talk. The company recently provided very strong financial guidance: Marvell expects **fiscal year 2027 revenue to exceed $11 billion**, representing year-over-year growth of approximately **30%+**; by **fiscal year 2028**, revenue is expected to approach **$15 billion**, representing nearly **40%** further growth. The company itself also identifies its data center business as its largest growth engine, with data center revenue in the last quarter already growing **21%** year-over-year to **$1.65 billion**. This indicates that Marvell isn't just telling an AI story now; it's getting real AI orders onto its financial statements. More crucially, money from the entire industry is still pouring in. Large tech companies' **spending on AI infrastructure in 2026 is projected to exceed $630 billion**. This means that even if the market occasionally worries about a capex pullback, real demand in reality continues to expand. For a company like Marvell that works on AI interconnect, photonics, and custom chips, this isn't just icing on the cake; it's the firmest source of orders for the next two to three years. Therefore, my judgment on this investment is straightforward: **NVIDIA isn't saving Marvell; it's incorporating it into the infrastructure layer of the next phase of the AI ecosystem.** The biggest advantage of this layer of business is that once a platform is formed, customer stickiness, technical barriers, and bargaining power all become stronger. What Marvell feared most in the past was being tied down by ASIC orders and single customers. Now, through the line of XConn, Celestial AI, and NVIDIA, it's moving from order-driven to platform-driven. This change is the root cause of the stock's re-rating. Of course, the market won't automatically give it the green light just because of one investment. The reason Marvell's valuation hasn't completely taken off yet is also simple: although the AI interconnect narrative is getting stronger, execution, customer onboarding, and capital expenditure cadence still need to be consistently delivered. But if we place NVIDIA's $2 billion move within the context of the entire AI industry chain, the signal it sends is very clear: **What's truly valuable isn't just computing power itself, but who can connect, amplify, and run that computing power stably.** This is precisely the position Marvell is occupying. If you had to summarize in one sentence: **NVIDIA's $2 billion isn't just a bet on Marvell; it's a bet on the interconnect dividend of the next phase of the AI era.** This money isn't buying a quarter's stock price; it's buying an ecosystem entry point for a longer cycle. $NVIDIA(NVDA.US) ### 相关股票 - [NVIDIA (NVDA.US)](https://longbridge.com/zh-CN/quote/NVDA.US.md) - [Marvell Tech (MRVL.US)](https://longbridge.com/zh-CN/quote/MRVL.US.md) - [GraniteShares 2x Long NVDA Daily ETF (NVDL.US)](https://longbridge.com/zh-CN/quote/NVDL.US.md) - [XL2CSOPNVDA (07788.HK)](https://longbridge.com/zh-CN/quote/07788.HK.md) - [XI2CSOPNVDA (07388.HK)](https://longbridge.com/zh-CN/quote/07388.HK.md) - [YieldMax NVDA Option Income Strategy ETF (NVDY.US)](https://longbridge.com/zh-CN/quote/NVDY.US.md) - [Direxion Daily NVDA Bear 1X ETF (NVDD.US)](https://longbridge.com/zh-CN/quote/NVDD.US.md) - [T-Rex 2X Long NVIDIA Daily Target ETF (NVDX.US)](https://longbridge.com/zh-CN/quote/NVDX.US.md) - [T-Rex 2X Inverse NVIDIA Daily Target ETF (NVDQ.US)](https://longbridge.com/zh-CN/quote/NVDQ.US.md)