Elon Musk: This Is AI's Limiting Factor - Brownstone Research

In this issue 01 Three New Earnings Reports Confirmed the AI Spending Boom Nick Rokke 02 Elon Musk Just Quantified the Memory Shortage Joe Withrow 03 The Stock Market’s New Cash Cow Clint Brewer Three New Earnings Reports Confirmed the AI Spending Boom Nick Rokke Senior Analyst Earlier this week, three major companies sent Wall Street the same message: The artificial intelligence infrastructure boom shows no sign of slowing. In fact, it’s getting bigger. CoreWeave (CRWV) and Nebius Group (NBIS) are two of the biggest “neoclouds.” These are specialized cloud providers dedicated to delivering enormous amounts of AI compute. Super Micro Computer (SMCI) sits one layer deeper in the infrastructure stack. It takes GPUs, networking equipment, cooling systems, and other components and assembles them into complete server racks ready for installation inside data centers. All three companies just reported explosive results. And together, they confirmed that demand for AI compute is still outrunning supply. CoreWeave’s second-quarter sales surged 112% year-over-year to $2.6 billion. And its adjusted EBITDA doubled to $1.5 billion. But most importantly, CoreWeave’s backlog jumped 246% to $104 billion. And that figure doesn’t even include the $25 billion in additional customer commitments signed early in the third quarter. Management responded by raising its full-year revenue outlook to $12.8 billion. It also increased expected capital expenditures (capex) to $37 billion. That’s a staggering amount of money. But management has said it has multiple customers interested in every GPU the company brings online. Even older chips remain in demand. CoreWeave signed a contract that will keep Nvidia A100 GPUs—first introduced in 2020—working through 2029. That tells us these systems can continue producing revenue long after newer generations arrive. Nebius reported even faster growth, albeit from a smaller base. Revenue surged 454% year-over-year. And adjusted EBITDA swung from a $21 million loss during the same quarter last year to a $236 million profit. Nebius also closed four major AI cloud agreements, each with a contract value exceeding $1 billion. And this is where the story gets more compelling. Those agreements included customer prepayments covering an estimated 50%-60% of the capital needed to build the dedicated infrastructure. Customers are helping finance new AI data centers because they want to guarantee access to scarce computing power. While this doesn’t eliminate the financing risk facing these neoclouds, it does address one of Wall Street’s biggest concerns—the fear that these companies will be unable to raise enough capital. Then came Supermicro. Quarterly revenue nearly doubled year-over-year to $11 billion. And the company also entered the new quarter with a record backlog. We are now seeing the same signal across the entire AI infrastructure chain. Semiconductor companies like Nvidia (NVDA) and AMD (AMD) continue raising estimates. Server companies like SMCI are receiving record orders. And hyperscalers like Amazon (AMZN) and Google (GOOG) are seeing increasing demand. And that’s flowing into the neoclouds. None of this makes every neocloud stock an automatic buy. Financing costs, execution risk, and customer concentration still matter. But we can clearly see demand is increasing. As AI moves from training models to running inference and autonomous agents, computing power becomes a recurring requirement. Every AI model placed into production consumes chips, memory, networking, cooling, and electricity around the clock. CoreWeave, Nebius, and Supermicro confirmed the AI infrastructure boom is still accelerating. Recommended Links “Market Wizard” Reveals Strategy to Help Americans Target $1,000 Larry Benedict, famous for handing his clients hundreds of millions of dollars in his career, is revealing his “two-rule” strategy to help his followers go for their first $1,000 in the days and weeks ahead. Join him for just $19.. Click here now. Jeff Brown Says "SpaceX Supercycle" Could 39x Your Money According to legendary tech investor Jeff Brown… Three words on page 37 of the SpaceX IPO filing… Signal Musk is about to trigger a rare wealth supercycle… And send billions into a sector nobody associates with tech. Click Here Now To Watch His Urgent Strategy Session Elon Musk Just Quantified the Memory Shortage Joe Withrow Senior Analyst Buried within SpaceX’s second-quarter 2026 earnings call, Elon Musk said something incredible that every investor should be aware of… and almost nobody noticed. Asked about rising infrastructure costs, Musk didn’t blame GPUs or energy. He blamed memory. Musk specifically singled out memory as the “limiting factor.” He also had this to say: But ask yourself, is the [memory] demand increasing by 20% a year? No, the demand is increasing by 200% a year, maybe higher. So if you’ve got demand increasing much faster than supply, then Economics 101 would suggest that the price increases. It does not decrease. When we think of SpaceX, we think of rockets and satellites and missions to the Moon, Mars, and beyond. And that’s why Musk’s key insight in the quarterly earnings call got very little attention. When talking about a rocket company, who wants to talk about boring old memory? But Musk’s view on memory has important implications. Along with power capacity, memory has become a bottleneck in the artificial intelligence (AI) arms race. That’s because memory — specifically high-bandwidth memory (HBM) stacked next to the GPUs — is what feeds the processors with the enormous volumes of data they need. Modern AI models contain hundreds of billions or even trillions of parameters. During both training and inference, those parameters must be read from memory at breakneck speed. So, without enough fast memory capacity and bandwidth, the most powerful GPUs in the world simply sit idle waiting for data. As Musk pointed out, memory supply growing at 20% a year would have been considered fast just a few years ago. But with the race for AI in full swing, memory demand is compounding at 200% per year. That means supply is falling an order of magnitude short of demand. And given that nobody can simply spin up a new memory fab on short notice, this is a supply gap that won’t be filled anytime soon. Musk’s own conclusion was blunt: “Economics 101 would suggest that the price increases.” This isn’t a stray comment from an outsider. It’s the CEO of a company burning $18.4 billion in quarterly capex on AI infrastructure, watching in real time as one of his own largest cost lines outpaces every model his team built. So, we can be confident that the cost of HBM and the systems that depend on it will keep rising. For investors, this signals that companies controlling scarce memory capacity, advanced packaging, or the ability to secure long-term HBM supply will capture outsized value in the next phase of the AI buildout. This is a big reason why Jeff originally recommended Micron (MU)—a giant in the memory space—to The Near Future Report subscribers in October 2024. Subscribers are already up more than 780% on that position. And if Musk’s comments are even close to accurate, Micron’s run could be far from over. The Stock Market’s New Cash Cow Clint Brewer Research Analyst, Opportunistic Trader For all the fuss around AI stocks, you likely overlooked the best-performing S&P 500 sector this year. The top spot doesn’t belong to technology. It’s energy. The energy sector is outpacing all others this year with a gain of 36%. The tech sector is a distant second with a 23% return. Energy stocks got off to a quick start in 2026. You can thank President Trump for that. First, it was a surprise military operation inside Venezuela that culminated in the capture of that country’s president. Venezuela is a large oil producer and sits on the biggest oil reserves in the world. Next came the war between the U.S. and Iran that’s still dragging on and stifling traffic through the Strait of Hormuz. Before the war, 20% of the world’s oil flowed through the Strait every day. Conflict in key oil-producing regions sent oil prices higher. Energy stocks followed along…until they went too high. Even as oil spiked over $110 per barrel in early April, energy stocks were already pulling back. Let me explain why. There’s a sweet spot for oil prices—one that maximizes cash flow and profits for energy companies without causing demand destruction. When oil prices rise, it can increase the margins for the companies pulling that oil out of the ground. But if it gets too high…it starts to pinch consumers. They put off that long road trip. Businesses that incur transportation costs raise prices on consumer goods, which can result in consumers forgoing those goods. That’s demand destruction in a nutshell. Many economists agree that the sweet spot is somewhere between $70 and $90 per barrel. As it happens, the TV show “Landman” explains this dynamic perfectly . The important thing is that oil has been trading in that sweet spot for the last two months, and the results are showing up in energy companies’ financials. Free cash flow yield is a way to measure how much cash a company is generating from its operations less capital expenditures relative to its market value. Collectively, energy stocks are expected to produce a free cash flow yield of 8.4% this year…the highest of any sector. That comes at a time when AI hyperscalers are plowing massive sums into capital expenditures to support the AI infrastructure buildout. Hyperscalers used to be the market’s cash cows. But they could end up spending as much as $860 billion in capex this year and $1.2 trillion in 2027. Free cash flow across eight hyperscalers is estimated to swing from +$180 billion in 2025 to -$144 billion in 2027. If oil prices stay in the sweet spot, then energy stocks are the stock market’s new cash cows. And if you missed the opportunity to ride energy stocks higher at the start of the year, you might get another chance. Here’s the chart of the State Street Energy Select Sector SPDR ETF (XLE). After a 38% rally to start the year, XLE hit a peak near the $62 level back in late March. But since then, the ETF has been trading in a consolidation pattern that could set the foundation for another advance. Since the top, XLE is forming a pattern that resembles an “inverse” head and shoulders pattern. You’ve probably heard of a head and shoulders chart pattern, which is usually a topping signal. The inverse version works the other way and will sometimes form in the middle of an advance. XLE is now testing the highs from March. If oil prices stay in the sweet spot, that could drive another leg higher in energy stocks.