Home Technology Nvidia mostly sat out the chip sector's best quarter...
Technology

Nvidia mostly sat out the chip sector's best quarter ever. What needs to change?

Key Points

Semiconductor stocks are about to complete their best quarter ever. And yet, the biggest chipmaker of them all — Nvidia — has largely sat out the rally. To reignite its stock, Jim Cramer said Nvidia needs to open its checkbook and return more cash to investors.

Semiconductor stocks are about to complete their best quarter ever. And yet, the biggest chipmaker of them all — Nvidia — has largely sat out the rally. To reignite its stock, Jim Cramer said Nvidia needs to open its checkbook and return more cash to investors. The Philadelphia Semiconductor Index , known as the SOX, has soared more than 80% in the second quarter, as demand for artificial intelligence computing strengthened and broadened. The SOX leaderboard for the quarter is dotted with blistering gains: Micron is up 239% through Monday's close, fueled by soaring prices for memory and storage chips thanks to a major supply crunch. Lam Research , which makes essential equipment used in the semiconductor manufacturing process, is up 92%. Club name Intel has nearly tripled, and Advanced Micro Devices has ripped more than 165%. Both companies are longtime makers of central processing units (CPUs) for data centers. The first wave of the AI compute boom was driven by accelerator chips, most notably Nvidia's graphics processing units (GPUs). The second quarter, however, marked Wall Street's full recognition that an emerging type of AI computing — agentic systems capable of completing tasks autonomously — requires a ton of CPU power to go along with the GPUs. This also helped a fellow Club name, Arm Holdings , which is pushing deeper into the CPU market, climb by over 125% during that period. Even Texas Instruments , long considered a boring bet on industrial markets, got in on the action; it's up some 47% thanks to accelerating growth for power management chips used in data centers. Then there's Nvidia, whose cutting-edge GPUs kickstarted this AI boom and turned it into the world's most valuable company. It is the worst-performing stock in the entire SOX in the April-to-June period, with a gain of roughly 12%. The stock's frustrating performance relative to its chip peers dates back to last year, but the second-quarter weakness hammers the point home. This has been a stretch in which investors seemingly wanted only to buy semiconductor stocks, yet Nvidia has been unable to regain the momentum of yesteryear. .SOX NVDA 1Y mountain The Philadelphia Semiconductor Index versus Nvidia's stock over the past 12 months. The problem cannot be explained by looking at Nvidia's reported results. In its May 20 earnings report, the annual growth rate of its data center business actually accelerated. The segment's revenue surged 92% in the April quarter to $75.2 billion, up from 75% growth to $62.3 billion in the February quarter. Plus, its guidance for the May-to-July period was comfortably above Wall Street estimates. What's ailing the stock is a combination of factors — some related to market dynamics and others to hard-to-quell concerns about rising competition. As the AI compute narrative evolved this year, Nvidia's stock was likely a source of funds for investors looking to capitalize on memory and CPU shortages, and the need to buy more chipmaking machines from the likes of Lam Research and Applied Materials . Nvidia shares rose 1,000% from the launch of ChatGPT in late November 2022 through 2025, compared with a 75% advance for the S & P 500 and a 159% gain for the SOX. In other words, there's been a ton of money already made in the stock, and investors — especially short-term traders like hedge funds — are always hunting for the hot new thing. As Jim has stressed before , companies selling a product in short supply are the hot new things in this market. Companies like Micron and Applied Materials are saying they don't know when supply and demand will catch up, Jim said Tuesday on CNBC. "Nvidia does not and cannot say that." Of course, we're also always looking for new ideas at the Club. It's why we added Arm Holdings to the portfolio in April and Intel earlier this month . Arm is a straightforward bet on CPU growth. Intel also offers greater exposure to the CPU market and could benefit as the industry seeks an alternative to Taiwan Semiconductor Manufacturing Co ., which dominates advanced chip production and packaging. Nvidia also makes CPUs, but the company is still best known for its bread-and-butter GPUs. Nvidia's GPUs and other AI accelerator chips perform the complex math that underpins AI models. In agentic AI systems, CPUs are responsible for orchestration and for keeping the GPU busy. But they also handle many of the tasks that agents are asked to do — like searching the web and updating records in a database. The two kinds of chips work together, but agentic systems require far more CPUs than a chatbot that gives a user a simple written answer to their question. This brings us to the next layer of Nvidia's stock rut: concerns about competition facing its GPU empire. And this is an overhang that Jim is especially worried about, as he made clear on Tuesday. The competition is coming from not just legacy chip designers such as AMD, which is the No. 2 supplier of data center GPUs. It's also emanating from many of Nvidia's biggest customers themselves. Companies such as Alphabet -owned Google, Amazon , Microsoft , and Meta Platforms all make their own AI chips. OpenAI, the ChatGPT creator, is also advancing its in-house silicon efforts, working closely with fellow Club name Broadcom . Google and Amazon's chips are generally considered the best of this bunch. Google has worked closely with Broadcom to co-design its tensor processing units (TPUs) for years. The TPU has been a key advantage in the AI race, and Google is taking steps to monetize the chips in new ways, including a cloud venture with Blackstone and allowing customers to buy TPUs for their own data centers. Amazon's Trainium accelerator has been used by Claude model maker Anthropic to great success, burnishing its reputation. Similarly, Amazon is exploring selling its Trainium chips to customers rather than solely renting access through the cloud, Bloomberg News reported earlier this month. One of the big advantages that Google and Amazon tout about their chips is efficiency, allowing them to charge less for computing power than Nvidia. "When you listen to Google, they're telling you, 'Listen, we've got a better product. You listen to Amazon [and they're saying], 'We've got a better product,'" Jim said. "We have major customers who are saying we can compete.... You start thinking, wait a second, maybe these companies are going to encroach on Nvidia's territory." As much faith and trust as we have in Nvidia CEO Jensen Huang's leadership and vision, that is a tough narrative to completely refute based on the current set of facts. This isn't the first time that the market has grappled with the idea of Nvidia facing competition and slower growth as a result. But it is arguably the loudest that these concerns have been during the AI boom. So, what's Nvidia to do to keep delivering on the important task of making shareholders money? Jim said the company must increase the amount of stock it repurchases. This is in addition to continued investment in research and development, which helps ensure its technology improves from generation to generation. Nvidia spent $13.65 billion on R & D in its fiscal year ended in January, up 45% from the prior year's $9.4 billion. But even with that step-up in spending, Nvidia's free cash flow totaled some $97 billion. Most importantly, it's projected to be over $200 billion this fiscal year. That's the money we want Nvidia to deploy into buybacks, which should help shrink its share count and increase earnings per share. That's the money we want Nvidia to deploy into buybacks, which should help shrink its share count and increase earnings per share — assuming that its net income at least holds steady. The ideal scenario is increasing net income and a smaller share count, which amplifies EPS growth. This matters because stocks are ultimately valued on earnings growth. Nvidia spent $40 billion on buybacks last fiscal year and $33.7 billion in the year ended January 2025. The company's board approved an additional $80 billion repurchase program in May, on top of the $38.5 billion remaining under an existing buyback authorization. Jim still argued that it's not enough. "Nvidia must do what Apple did and buy back an endless amount of stock," Jim said Tuesday. While this isn't the first time Jim has implored Nvidia to take a page from Apple's playbook, the rationale for his argument hasn't changed. Under the leadership of outgoing CEO Tim Cook and ex-CFO Luca Maestri, who retired in 2024, Apple adopted an aggressive repurchase program . It started with almost $23 billion worth of repurchases in the 12 months ended September 2013, according to FactSet, and scaled up from there as the iPhone maker's business and cash flow grew. Apple spent $95 billion on buybacks in fiscal 2024 and almost $91 billion in fiscal 2025. Working in Nvidia's favor is that the company wouldn't be spending its money on an overpriced stock. Nvidia's forward price-to-earnings multiple has compressed so much that it's now cheaper than the broader market. Nvidia trades at roughly 19 times forward earnings versus 20.4 for the S & P 500, FactSet data shows. Nvidia's five-year average is roughly 36, according to FactSet. "I say Nvidia 'own it, don't trade it.' I'm not backing away from that," Jim said. However, he added, "Nvidia has to buy back stock hand over fist." (Jim Cramer's Charitable Trust is long NVDA, INTC, ARM, AVGO, GOOGL, META, MSFT. See here for a full list of the stocks.) As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after sending a trade alert before buying or selling a stock in his charitable trust's portfolio. If Jim has talked about a stock on CNBC TV, he waits 72 hours after issuing the trade alert before executing the trade. THE ABOVE INVESTING CLUB INFORMATION IS SUBJECT TO OUR TERMS AND CONDITIONS AND PRIVACY POLICY , TOGETHER WITH OUR DISCLAIMER . NO FIDUCIARY OBLIGATION OR DUTY EXISTS, OR IS CREATED, BY VIRTUE OF YOUR RECEIPT OF ANY INFORMATION PROVIDED IN CONNECTION WITH THE INVESTING CLUB. NO SPECIFIC OUTCOME OR PROFIT IS GUARANTEED.
Nvidia (ORG) Jim Cramer (PERSON) The Philadelphia Semiconductor (ORG) SOX (ORG) Micron (ORG) Lam Research (ORG) Intel (ORG) Advanced Micro Devices (ORG) AI (ORG) CPU (ORG) Club (ORG) Arm Holdings (ORG) Texas Instruments (ORG) Philadelphia Semiconductor (ORG)
Originally published by CNBC Read original →