Arm Holdings CEO Rene Haas on Thursday sought to reassure investors about one of the biggest sticking points after its earnings report: Will Arm be able to produce enough of its AI chips to meet demand? In an interview with Jim Cramer on CNBC, Haas said he was “confident” that Arm would secure a sufficient supply of its new central processing units (CPUs) to meet its stated $2 billion in customer demand across fiscal 2027 and fiscal 2028. Arm’s fiscal 2027 started last month. “This is not perishable demand, to be clear,” Haas said. “This is demand that is firm, sustaining, and very, very robust. Agentic AI puts a huge amount of pressure on the CPU to do all the work around orchestration, scheduling, and the management of these agents. That’s only work the CPU can do. So, while we are in the process of securing supply for that additional demand, the demand is not going away. I’m confident we’ll get that supply, and I’m also very confident that demand is going to continue.” The reason why this matters: That $2 billion figure is double what Arm laid out in March when it announced the chip, branded as the AGI CPU. Investors were excited to see that upward revision in Arm’s shareholder letter released after Wednesday night’s close, and it helped explain why the stock got an after-hours pop. Doubts arose, though, when CFO Jason Child said on the earnings call about an hour later that Arm was maintaining its official outlook of $1 billion in AGI CPU revenue over the next two fiscal years “while we pursue supply chain capacity.” He added, “We still expect the first revenues from production chip sales to land in the fourth quarter of this fiscal year.” Shares started to lose steam on the supply chain comment. Now layer in that the stock was red hot coming into the print, and it’s unsurprising to see shares down 10% on Thursday. Arm shares ended Wednesday up about 75% since debuting the AGI CPU on March 24. Under these conditions, talking about line-of-sight into more revenue but not actually upping the forecast is going to invite profit-taking. On Wednesday, Jim warned that the stock was vulnerable to a pullback at any hint of imperfection. ARM 1Y mountain Arm’s stock performance over the past 12 months. It turned out the imperfection was around the supply chain, an incredibly complex network of designers, suppliers, and manufacturers all scrambling to meet soaring demand for computing power, as companies spend hundreds of billions of dollars to build new data centers stocked with server racks. Among the biggest bottlenecks right now is securing enough capacity at Taiwan Semiconductor Manufacturing Co. , the world’s most advanced third-party chipmaker, to have your product made and packaged. Even though TSMC is ramping up its capacity, volume production cannot happen overnight due to the sheer complexity of the task. Nvidia , Advanced Micro Devices , and Broadcom (co-designer of Google’s popular in-house AI chip) are all major customers of TSMC. Amazon’s custom Trainium chips are made at TSMC , too. They’ve worked to secure their place in line. Even Apple , long believed to be TSMC’s biggest customer , said last week it’s facing supply constraints on its compute systems called systems-on-a-chip (SoCs). In other words, Arm isn’t alone in racing to secure supply. But the crowded field amplifies concerns about whether they will secure enough. “These chips take a while to design,” Haas said. “They take a while to build. They take a while to get through the entire pipeline,” Haas said. “So we are working with our partners in TSMC for months, if not years, to make them aware of the product, the supply, what we think the demands look like, what we think the forecast look like. We’re talking to them all the time. So, it’s not a surprise. It’s not like we whipped up to TSMC and said, ‘Hey, look, we’ve got this chip we announced on March 24. Can you help us?’ All of that is lined up.” Arm’s business model has historically been centered on licensing its intellectual property — its Arm “instruction set” that enables hardware and software to communicate, and blueprints for other parts of chips — to external customers. The AGI CPU represents its first foray into making the entire chip itself. Wall Street analysts would frequently ask Arm about reports that it was planning to make a complete chip, Haas noted, but the company was reluctant to provide details. “We really only wanted to talk about it when the product was back from TSMC. It was working, the customer validated it, and we were ready for production,” Haas said, and now those milestones have been hit. “All that work, that was all happening in the background. It has to.” Bottom line We’re happy that Haas took these questions head-on and expressed confidence in the company’s ability to navigate a tight supply chain. Execution risk is no doubt a real concern whenever a company embarks on a new product, and it will be something to monitor regularly. At the same time, demand for computing power is off the charts, especially for CPUs, because of the advent of agentic AI systems capable of executing tasks without human intervention. These systems don’t just answer a question with text or an image. They can go out and complete tasks such as booking flights and managing inventory. In the first couple of years of the AI boom, demand was strongest for graphics processing units (GPUs) from companies like Nvidia, which helped train massive models. CPUs were an afterthought. Now, CPUs are increasingly essential to the day-to-day use of the models, known as inference. That trend isn’t slowing down. Nvidia’s CPUs are built on the Arm instruction set. The same goes for Google and Amazon’s CPUs. Arm-based CPUs are known for their power efficiency, so these companies pair them with their power-hungry AI workhorse chips, such as GPUs. This gives us confidence that Arm’s business collecting royalties will continue to thrive as its in-house chip efforts scale. In more ways than one, Arm is in the sweet spot to reap the rewards of the AI boom. (Jim Cramer’s Charitable Trust is long ARM, NVDA, AAPL, AMZN, and AVGO. 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.
Arm CEO looks to set record straight after stock’s post-earnings fall
Posted by
No Comments
No Comments
Post navigation
Previous Post
WHO: Hantavirus-Ausbruch nicht der Beginn einer Pandemie