Arm Holdings shares fell Wednesday night even as the chip designer reported a better-than-expected quarter and gave a positive outlook for its data center CPU business. The company’s fiscal 2026 fourth-quarter sales (ending March 31) rose 20% year-over-year to $1.49 billion, beating the consensus estimate of $1.47 billion from analysts compiled by LSEG. Non-GAAP earnings per share (EPS) increased 9% to 60 cents, beating expectations of 58 cents. ARM YTD Mountain Arm Holdings YTD Arm stock fell about 6% in after-hours trading, recouping about half of its gains during regular trading. We raised the possibility of this happening during Wednesday’s club member morning meeting. There is a possibility of a drop in inventory due to large numbers and last-minute rush before printing. That’s exactly what happened. The stock closed at a record high of $237, pushing its year-to-date gain to 117%. Conclusion When I initiated a position in Arm last month at around $170 per share, I wanted to ensure my portfolio was exposed to the data center CPU market. The artificial intelligence revolution has evolved significantly over the past six months. Initially, it was all about the best graphics processing units (GPUs) for training large language models. The focus then shifted to inference, and now those workloads are evolving again, from processing human-generated prompts to supporting ongoing agent-driven tasks. While GPUs still play an important role in the future of AI, the once-obsolete central processing unit (CPU) is undergoing a major transformation. This CPU renaissance was confirmed when Intel reported on it two weeks ago. Intel CEO Lip Bu Tan said in an April 23 earnings call that the ratio of CPUs to GPUs in AI racks used to be 1:8. But with the rise of agents, it’s closer to 1:4 and could become parity, meaning 1:1, in the future. In other words, we need far more CPUs than we did a few years ago. Advanced Micro Devices gave a similar story during its earnings call Tuesday night. Quantifying how big the CPU market is getting, AMD CEO Lisa Su said she expects the total addressable market for CPU servers to grow at more than 35% annually, reaching more than $120 billion by 2030. In an interview with Jim on CNBC on Wednesday, Hsu said, “Agents are really driving huge demand throughout the AI adoption cycle.” It’s hard to see a stock price triple on the same information, so it wouldn’t be surprising to see Arm return some of its recent parabolic gains. However, we thought our case was solidified in the post-earnings conference call. Arm-based CPUs command over 50% share among top hyperscalers. While AMD and Intel may claim they have the market share advantage, Arm pointed out on the conference call that the three largest providers of AI accelerators are combining their chips with Arm-based chips. Nvidia’s Rubin GPU is integrated with the Vera (Arm-based) CPU. Google has Tensor Processing Units (TPUs) with Axion (Arm-based) CPUs, and Amazon has Trainium with Graviton (Arm-based) chips. All three are also portfolio names. “Whether it’s Nvidia, Amazon, Google, the very largest and most popular accelerators by volume are TPU, Trainium, and Rubin. … They’re all connected to Arm,” CEO Renee Haas explained on a conference call. Alphabet’s Google TPU is co-designed with fellow club name Broadcom. Reasons to own one Chip designer Arm is at the center of the CPU resurgence. The shift from AI training to model execution has reignited the demand for central processing units. Arm has a lucrative licensing and royalty business for its chip architecture, which is widely used by major hyperscalers. In March 2026, Arm announced the next chapter in its story: its first in-house data center CPU designed specifically for agent AI workloads. Competitors: Advanced Micro Devices, Intel Recent Purchases: April 20, 2026 Start Date: April 20, 2026 AI giants are increasingly favoring Arm-based CPUs over traditional x86 processors, an architecture dominated by AMD and Intel, due to their performance benefits and increased efficiency. Arm’s business model has traditionally focused on collecting upfront licensing fees and royalties on chip shipments, but the new element in the story is developing its own chips. Customer response to ARM AGI CPUs seems to be great. When announcing its first-ever in-house data center CPU at the Arm Everywhere event in March, the company said it expected demand of more than $1 billion over the next two years. It’s been less than two months and management is already doubling down on this view. We currently expect customer demand to exceed $2 billion from the end of fiscal year 2027 to the end of fiscal year 2028. However, it tempered its positive outlook slightly, saying it was maintaining its original $1 billion outlook as it needed to adjust supply chain capacity to meet demand. Concerns about these supply constraints caused stocks to give up initial gains after hours. As I mentioned when I first added Arm to my portfolio, the company has a great pitch when it comes to CPUs. The company believes hyperscalers have the potential to reduce AI data center capital expenditures (capex) by up to $10 billion per gigawatt. That’s all given the market’s focus on free cash flow. The long-term goal remains $15 billion by the end of fiscal 2031, and these sales are not expected to cannibalize Arm’s existing business, a key factor pushing back on bearish theories. “The main reason we did this is because our customers asked for it. At the end of the day, we’re responding to customer demand in the market,” Haas said of developing his chip. The bottom line is that demand for Arm-based data center CPUs is unexpected, supporting strong double-digit revenue growth for the foreseeable future. The success of homegrown chips makes the story even better, and it’s up to management to navigate the tight and complex supply chain environment to achieve their goals. Given the stock’s recent parabolic movement, we maintain a $250 price target and a Hold rating of 2. In the short time since Arm was added to the portfolio, the stock has increased nearly 40% as of Wednesday’s close. If out-of-hours transfers are valid, we will refund a portion of your advance payment. But the gathering in the Arm was the same as our April 20th launch, and for that matter, it was incredible in 2026. Commentary Regarding quarterly results, license and other revenues increased approximately 29% year-over-year to $819 million, exceeding market expectations. These revenue streams come from upfront license fees collected from customers who want access to the company’s CPU architectures and designs. Royalty revenue rose 11% year over year to $671 million, but was actually lower than street expectations. However, this shortage is likely due to the smartphone market. The business is still growing year over year, but the end market is depressed due to memory shortages. More importantly, adoption of Arm-based server chips by all major hyperscalers is accelerating and adoption of data center networking chips is increasing. We’re also pleased that Arm’s gross margin and operating margin exceeded expectations. Arm’s current revenue streams are all licenses and royalties, generating very attractive gross margins. On a non-GAAP basis, it was 98.32% for the quarter. (GAAP stands for Generally Accepted Accounting Principles. Non-GAAP, also known as adjusted, removes temporary elements in hopes of achieving an apples-to-apples comparison on a quarterly basis.) Non-GAAP operating margins were also better than expected, with cost growth down from a 26% compound annual growth rate (CAGR) from 2024 to 2026. Operating leverage should further increase in the future as it slows down to mid-%. CAGR from FY2026 to FY2031. Outlook Arm provides guidance on a quarterly basis. For the first quarter of 2027, the company expects revenue to be $1.26 billion plus or minus $50 million, or in the range of $1.255 billion to $1.265 billion. According to LSEG, this is slightly above the consensus estimate of $1.25 billion. (However, this will be down sequentially, as the fiscal fourth quarter was $1.49 billion.) The company expects non-GAAP operating expenses to be $760 million, slightly higher than the FactSet consensus estimate of $742 million. Non-GAAP earnings per share are expected to be 40 cents, plus or minus 4 cents, or a range of 36 cents to 44 cents. That beats the consensus estimate of 36 cents, according to LSEG. (Jim Cramer’s charitable trusts are long ARM, NVDA, GOOGL, AMZN. See here for a complete list of stocks.) As a subscriber to Jim Cramer’s CNBC Investment Club, you will receive trade alerts before Jim makes a trade. After Jim sends a trade alert, he waits 45 minutes before buying or selling stocks in his charitable trust’s portfolio. If Jim talks about a stock on CNBC TV, he will issue a trade alert and then wait 72 hours before executing the trade. The above investment club information is subject to our Terms of Use and Privacy Policy, along with our disclaimer. No fiduciary duties or obligations exist or arise from your receipt of information provided in connection with the Investment Club. No specific results or benefits are guaranteed.
