Here is why Nvidia’s partnership and investment in Marvell is such a big deal
Nvidia’s strategic tie-up with chipmaker Marvell Technology is another reason why the AI giant is hanging on to its stagnant stock. On Tuesday morning, the companies announced a partnership to integrate Marvell’s custom AI chips and networking products with Nvidia’s dominant AI computing ecosystem. Also, club name Nvidia is buying a $2 billion stake in Marvell, its latest investment in other tech companies, using its cash windfall from the AI boom. For Nvidia investors, the deal is important because it strengthens Nvidia’s presence in the world of custom AI chips, which are widely considered the biggest threat to its throne as the king of AI computing. Nvidia’s biggest customers are still buying staggering amounts of computing from Nvidia while investing in their own custom silicon with design help from Marvell and fellow Club owner Broadcom. While Marvell’s customers include Amazon and Microsoft, Alphabet and more recently Meta are known to be working with Broadcom. Nvidia shares rose 5.6% on Tuesday, their second best day of the year. It’s an encouraging reaction after months of the stock weakening despite a steady stream of good news; a prime example of this is the frustratingly subdued market reaction to Nvidia’s impressive GTC conference in mid-March. Undoubtedly, some of Nvidia’s recent weakness stems from the broader market pullback due to the Iran war. But whatever the reason, the result was that the stock hasn’t been this cheap in more than a decade. Now with the Marvell arrangement, we have even more reason to believe Nvidia will continue to deliver earnings growth in the coming years. In fact, Nvidia will have an opportunity to capture a larger share of the money spent on AI data centers if its other products are more compatible with custom silicon. Although Nvidia is best known for its powerful graphics processing units (GPUs), it has become a one-stop shop for AI computing, offering additional non-GPU types of processors and networking technology that allows all the chips to talk to each other. Nvidia also has a large software library driven by its CUDA (Compute Unified Device Architecture) platform, which forms the basis of its competitive position. The deal with Marvell focuses on the networking aspect of its portfolio, specifically its NVLink technology. Last year, Nvidia announced the creation of NVLink Fusion to effectively open its proprietary structure to non-Nvidia processors, and that’s what the companies will benefit from in this strategic partnership. This news may have been foreshadowed in December when we learned that Amazon would integrate NVLink Fusion with its own custom silicon solutions in cloud computing, including the Tranium4 chip and Graviton central processors (CPUs). In an interview on CNBC on Tuesday morning, Nvidia CEO Jensen Huang explained the reasoning behind the Marvell alliance. “All the data centers of the world will be replaced by this new way of computing,” Jensen told Jim Cramer. “Of course, the vast majority of data centers will be powered by Nvidia CUDA GPUs. However, for our customers who also want to have custom versions they build, we are expanding our architecture, starting with the network architecture, basically the entire chassis of our Grace Blackwell and Vera Rubin systems. We will extend that through NVLink and connect it to Marvell. Together, we will be able to ask our customers if they want to use all their Nvidia hardware or if they want to augment it. Our Nvidia hardware with their custom processors.” “Together we will address a much bigger problem” [total addressable market, or TAM]” said Jensen. Grace Blackwell is Nvidia’s current-generation computing platform powered by Grace GPUs and Blackwell GPUs. It is on track to launch Vera CPUs and Rubin GPUs — under the Vera Rubin family of server racks — toward the end of this year. Jensen said a company designing its own custom AI chip with NVLink Fusion can more easily connect them to Nvidia products. He specifically highlighted the SpectrumX networking framework as well as the new Vera CPUs. Daily AI model usage The Groq-powered inference chip and data storage platform known as Bluefield, designed for the Groq-powered chip and an updated version of Bluefield, were among the bullish updates introduced at GTC. “All of these will be compatible with the system architecture. The ecosystem will be compatible… [and] “It makes it easier for them to work together and great for us as we expand TAM,” Jensen said. Whether companies buy GPU compute platforms from Nvidia or invest in custom silicon and try to partner with Marvell to do it, Nvidia will benefit. Also keep in mind: the $2 billion investment means Nvidia will also benefit from any Marvell deals that don’t involve any Nvidia components, given that it represents a roughly 2.5% ownership stake in the company. Nvidia has also used its huge cash pile to invest in startups such as ChatGPT creator OpenAI and vendors both involved in the optical technology space. As AI computing intensifies, optics are becoming an increasing focus for Nvidia, and Tuesday’s announcement also noted it as an area of development. Its underperformance is even more surprising and its valuation even more shocking. Heading into Tuesday’s session, Nvidia’s shares were trading at about 20 times forward earnings, according to FactSet data — about as cheap as they were two years ago, when the stock’s forward price-to-earnings ratio was about 36. The P/E ratio was around 34 in August 2025, when shares started to stagnate. Over the years, we’ve only gotten close to the current P/E ratio of 20 twice, as a result of the Federal Reserve’s market-shattering tantrum and in the days following President Donald Trump’s April 2025 reciprocal tariff announcement. Crucially, Nvidia’s current valuation occurred before any uptick in earnings associated with the Marvell partnership emerged. So, time will tell how much faster the initiative accelerates. Profitable growth for Nvidia this year, next year and beyond, but the point is, the stock may be cheaper than we thought on Monday night. As a result, despite Tuesday’s rally, the stock is still below where the Iran war began. Of course, this could become a value trap where earnings keep rising but no one cares, but we don’t think that will ever be the case. Given what we heard at GTC and what we learned on Tuesday, we’re hard-pressed to see how high the estimates are, barring any dramatic intensification in Iran that sends the economy into recession and customers are forced to hold back on their AI spending. Perhaps Tuesday’s rally could be the beginning of that (although we’re not quite ready to ring the bell). No matter how good the news is, it’s hard to trust a name that doesn’t look like it can move higher until Tuesday. But we absolutely shout that Nvidia has given you one more reason to own the stock for the long term and not try to trade the name for every headline. Trust is long NVDA, AVGO, AMZN, MSFT, META, AVGO, GLW and GOOGL See here for a full list of stocks.) When you subscribe to the CNBC Investing Club with Jim Cramer, after Jim sends a trade alert, he waits 45 minutes before buying or selling a stock in his charitable foundation’s portfolio. THE ABOVE INVESTMENT CLUB INFORMATION, WITH THE ABOVE INVESTMENT CLUB INFORMATION, DISCLAIMERS OF LIABILITY, NO OBLIGATION OR DUTY TO SATISFY YOU EXISTS OR SHALL BE CREATED BY YOUR RECEIPT OF ANY INFORMATION PROVIDED IN THIS REGARD. CONNECTION WITH THE INVESTMENT CLUB DOES NOT GUARANTEE A SPECIFIC RESULT OR PROFIT.
