GPUs Built the Boom, But the Next AI Trade Isn’t What You Think VIEW IN BROWSER
Where will the next big gains in AI stocks come from? That’s the question every investor is trying to answer right now, and some of the most important clues are emerging this week. Because the center of the AI universe this week isn’t Wall Street. It’s the Nvidia GTC 2026 conference in Silicon Valley. Every year, thousands of engineers, developers and technology executives gather to hear Nvidia Corp. (NVDA) CEO Jensen Huang outline what’s coming next in AI. But GTC has become something much bigger than a developer conference. It’s where the AI industry telegraphs its next major moves. Investors who recognize the signals early are often the ones who capture the biggest gains. And this year, one signal is coming through loud and clear. While Nvidia built its empire on graphics processing units (GPUs), the company is shifting its focus to something far less glamorous, but potentially more important. The humble central processing unit, or CPU. At first glance, that may not sound exciting. But in the AI economy, it may mark the beginning of the next major shift in computing infrastructure… and where the next wave of investment gains may emerge. The Chip That Built the AI Boom For the past several years, GPUs have been the undisputed kings of artificial intelligence. These specialized chips can perform thousands of calculations simultaneously — making them ideal for training and running large AI models. That advantage created one of the most powerful supply squeezes the semiconductor industry has ever seen. Tech giants scrambled to secure GPU supply. Data centers expanded at a breakneck pace. And Nvidia became the single biggest beneficiary of the AI boom. In its most recent quarter alone, Nvidia generated more than $60 billion in data-center revenue, up roughly 75% from the year before. The company’s market value surged past $4 trillion, making it one of the most valuable businesses ever created. But the next phase of AI may require a different kind of computing power… The Rise of “Agentic” AI Until recently, most AI applications looked something like ChatGPT. You asked a question. The system generated an answer. But the next generation of AI is already moving beyond simple chatbots. Technology companies are now building agentic AI systems — networks of AI agents that collaborate to complete complex tasks. These systems can retrieve information, analyze data, make decisions, and coordinate with other agents in real time. Instead of answering one question… They manage entire workflows. That shift dramatically changes the computing demands inside data centers. GPUs remain essential for training and running AI models. But when multiple AI agents are coordinating tasks and moving large amounts of data across systems, general-purpose computing power becomes critical. That’s where CPUs suddenly matter again. Unlike GPUs — which contain thousands of small cores designed for parallel calculations — CPUs rely on a smaller number of powerful cores optimized for sequential processing. That makes them ideal for managing the orchestration layer of complex AI systems. In simple terms: The GPUs run the AI models… But the CPUs increasingly run the system that manages the AI. Nvidia’s Quiet Pivot Nvidia appears to recognize this shift. Several years ago, the company introduced its first data-center CPU platform, known as Grace. Now, the next generation — called Vera — is moving toward broader deployment. Recently, Nvidia struck a multiyear agreement with Meta Platforms Inc. (META) that includes large-scale deployment of Grace CPUs inside the social-media company’s data centers. Meanwhile, Nvidia-powered supercomputers using its CPU technology are already operating at institutions like Los Alamos National Laboratory and the Texas Advanced Computing Center. Why does this matter? Because the explosive demand for GPUs has exposed a new constraint in the system. CPUs are increasingly becoming the bottleneck when it comes to expanding AI and agentic workloads. Think about it this way. A modern AI data center might contain racks filled with extraordinarily expensive GPUs. But if the CPUs feeding those GPUs cannot move data quickly enough… Those GPUs sit idle. And idle GPUs are the last thing hyperscale data center operators want to see. A Quiet Supply Crunch The semiconductor industry is already seeing early signs of tightening supply for server CPUs. Delivery times for some processors have stretched toward six months. Prices have risen more than 10% in certain markets. Executives at Advanced Micro Devices Inc. (AMD) have described the surge in demand for data-center processors as “unprecedented” over the past several quarters. Intel Corp. (INTC) has also warned that its inventory levels could fall to unusually low levels before supply begins to recover. The problem is simple. Semiconductor manufacturing capacity cannot expand overnight. Building new fabs takes years. Increasing wafer production takes time. As one industry analyst recently put it: “Wafers don’t grow on trees.” That’s why some analysts now expect the global CPU market to more than double in size — rising from roughly $27 billion today to about $60 billion by 2030. But CPUs themselves are only part of the story. Because the deeper issue is something even bigger. | Recommended Link | | | | Luke Lango believes an entire sector is on the verge of a breakout, catalyzed by the largest government stock buying spree in a generation. He’s sharing every stock on his list, free of charge. Click here to get the full list. |  | | The AI Infrastructure Behind the Boom Right now, most investors are focused on the most visible winners of the AI Revolution. Companies like Nvidia dominate (see the headlines out of GTC this week). But tech revolutions rarely unfold in a straight line. They run into bottlenecks. And those bottlenecks often become the most profitable investment opportunities of all. I’ve seen this dynamic play out before. During the early days of the internet boom, demand for the raw materials used in electronics surged dramatically. Several mining companies delivered enormous gains as supply struggled to keep up with demand. During that time, I recommended companies like Antofagasta plc (ANTO.L), Freeport-McMoRan Inc. (FCX), Cameco Corp. (CCJ), and Impala Platinum Holdings (IMPUY) — firms supplying the raw materials behind the internet buildout. As that bottleneck tightened, those stocks delivered gains of 200%, 600%… even more than 800%, while the broader market went nowhere. Today, the AI Revolution is entering a similar phase. Because artificial intelligence doesn’t just require software. It requires physical infrastructure. - Chips
- Energy
- Memory
- Data centers
And in several of these areas, supply is struggling to keep pace with demand. That’s why I believe the next stage of the AI Revolution may be defined less by software breakthroughs and more by shortages. Preparing for the Next Phase of AI Because when a technological shift creates new bottlenecks, capital doesn’t stay where it is. It moves — often quickly — toward the companies positioned to solve them. That’s exactly the shift I break down in my brand-new free presentation, FutureProof 2026. During this event, I explain why several critical supply constraints are forming across the AI economy — and how investors can position themselves before those shortages become widely recognized. I also walk through several companies — including their names and tickers — that I believe could benefit as the next wave of AI infrastructure investment begins. You can watch it here now. Sincerely, | 
Post a Comment
Post a Comment