Goldman Says the AI Trade Isn't Over. The Market Is Voting With Its Feet.

The screens in the trading room were bleeding red. Not the deep crimson of a crash, but the anxious, flickering pink of a market losing its nerve. I watched the AI basket—the one everyone had been piling into since January—shed 10% in five days. The high-beta momentum fund, the one my buddy at a family office in Monterrey swore was bulletproof, was down 12%. You could feel it in the air, that specific kind of tension that only exists when leverage meets uncertainty. It was August, and the party that had been raging in AI stocks was suddenly looking for the exits. I pulled up the Goldman note that had been circulating through my WhatsApp groups, the one from their prime brokerage desk. The headline was a classic piece of Wall Street doublespeak: 'The AI Trade Is Not Over.' But the data inside told a different, more nuanced story. The rotation was real. Software had overtaken semiconductors as the largest weight in the three-month momentum long basket. Semiconductors and the AI complex—the very heart of the narrative—had been shoved into the short basket. This wasn't a pause. This was a reallocation. Let's rewind for a second. For the past eighteen months, the market has been operating on a simple, almost childlike logic: AI equals Nvidia, Nvidia equals the entire market. You didn't need to think about storage, or power, or even software. You just bought the chipmaker and watched your P&L go up. It was the easiest trade in the world, and everyone from the macro hedge funds in New York to the crypto degens in CDMX was on it. But that phase, as Goldman correctly points out, is changing. The era of getting paid just for being in the sector is over. Now, the market is demanding to see the receipts. This is where it gets interesting for someone like me, who spends his days looking at the plumbing of global liquidity. The Goldman note isn't really about AI. It's about the lifecycle of a narrative. Every asset class goes through this: the discovery phase, the mania phase, the distribution phase, and finally, the differentiation phase. We saw it with ICOs in 2017, with DeFi in 2020, and with NFTs in 2021. The first wave is always about the story. The second wave is about the fundamentals. The AI trade is now firmly in the second wave, and the market is doing what it always does during this transition: it's getting picky. Goldman's specific recommendation is telling. They're pointing investors toward storage and data centers, arguing that the 'profit recovery' in these areas hasn't been fully priced into the stock prices. This is a classic value-plus-catalyst play. The logic is sound: you can't run AI without massive amounts of memory and physical infrastructure. Nvidia sells the brains, but someone has to build the body. Micron, Dell, Super Micro—these are the picks-and-shovels plays of the AI gold rush. The market got so fixated on the shiny GPU that it forgot about the boring stuff that makes it all work. But here's where my contrarian instincts start to kick in. I've seen this movie before. In 2020, during DeFi Summer, the same thing happened. Everyone was chasing the yield on Uniswap and Yearn, ignoring the fact that the underlying protocols were mostly subsidizing their TVL with token emissions. The moment the incentives stopped, the users vanished. The 'profit recovery' in those protocols was a mirage. I can't help but wonder if the same dynamic is at play in the storage and data center trade. Is the demand for HBM and enterprise SSD real, or is it a function of hyperscalers front-loading their capex in a panic to keep up with the AI narrative? Let's look at the numbers more closely. Goldman mentions that the AI complex has been through a 'violent deleveraging.' The 10% drop in the AI basket and the 12% drop in the high-beta momentum basket are not small moves. They represent a significant unwinding of crowded positions. This is the market's way of saying that the risk/reward has shifted. When a trade is this crowded, any hint of bad news—a missed earnings estimate, a delay in a product launch, a hawkish comment from the Fed—can trigger a cascade. The question is whether this deleveraging is a healthy correction or the beginning of a more serious repricing. My read on the macro situation suggests we're in a delicate spot. The Fed is trying to thread the needle between fighting inflation and avoiding a recession. Liquidity conditions are tightening, and the M2 money supply, while still growing, is doing so at a much slower pace than during the pandemic. This is the environment where high-multiple, long-duration assets—which is exactly what AI stocks are—come under the most pressure. The 'risk-on' trade that fueled the AI rally was built on a foundation of zero interest rates and quantitative easing. That foundation is gone. The market is now operating on a different set of rules. This brings me to the contrarian angle that I think most people are missing. Goldman's report, for all its data, is still a sell-side document. It's designed to generate trading flow. The recommendation to buy storage and data centers is essentially a recommendation to buy the laggards of the AI trade. But what if the laggards are lagging for a reason? What if the 'profit recovery' in storage is already priced in, and the market is just waiting for the next shoe to drop? I've seen this happen with Layer 2 scaling solutions in crypto. For two years, the narrative was that 'decentralized sequencers' were just around the corner. The market kept buying the tokens, waiting for the technology to catch up with the PowerPoint. It never did. The same could be true for the storage trade. Let's talk about the elephant in the room: Nvidia's earnings. Goldman flags the Q2 report and the September industry conferences as the key catalysts. This is the moment of truth. If Nvidia beats expectations and raises guidance, the AI trade could reignite. If they miss, or if they give cautious guidance about the sustainability of demand, we could see a second wave of deleveraging that would drag down the entire complex, including the storage and data center names that Goldman is recommending. It's a binary event, and the market is pricing in a lot of uncertainty. I remember a similar moment in 2021 when I was heavily involved in the NFT market. I had bought three Bored Apes and a bunch of other PFPs, thinking I was building a digital art collection. The market was euphoric, and everyone was talking about the 'metaverse' and 'digital ownership.' But when the Fed started hiking rates, the liquidity dried up, and the floor prices collapsed. My $45,000 investment dropped by 60%. The lesson I learned was brutal: when the macro tide goes out, it takes everything with it, regardless of the quality of the underlying asset. The same principle applies here. If the Fed is forced to keep rates higher for longer, the AI trade—and the storage trade—will suffer, no matter how good the fundamentals look. So, what's the play? I'm not saying you should short the AI complex. That's a dangerous game, especially with Nvidia's earnings on the horizon. But I am saying that the era of blind buying is over. The market is entering a phase where differentiation is key. You need to be selective. You need to look at the balance sheets, the cash flows, and the actual earnings power of these companies. The 'profit recovery' that Goldman is talking about is real, but it's not going to be uniform. Some companies will thrive; others will be exposed as pretenders. Let's also consider the capital flows that Goldman mentions. The fact that money is rotating into European and Japanese banks, gold miners, and copper stocks is a significant signal. This is not just a rotation within the AI sector; it's a rotation out of the AI sector into other parts of the market. This suggests that investors are looking for value outside of the tech complex. It could be a defensive move, a hedge against a potential AI bubble. Or it could be a recognition that the AI trade has become too crowded and that better opportunities exist elsewhere. Either way, it's a sign that the market is broadening out, and that's usually a healthy development. The copper angle is particularly interesting to me. The fact that copper miners are being mentioned in the same breath as AI is a testament to the physical reality of the AI buildout. Data centers consume enormous amounts of electricity, and they require massive amounts of copper for wiring and cooling. This is a tangible, real-world demand that is not going away. It's a bit like the gold rush in the 19th century—the people who got rich were not the miners, but the ones selling the picks, shovels, and blue jeans. In the AI gold rush, the picks and shovels are not just GPUs; they're also the power infrastructure, the cooling systems, and the raw materials. But here's the thing that keeps me up at night: the timeline. Goldman is talking about a 'profit recovery' in storage and data centers, but they don't specify when this recovery will materialize. Is it in the next quarter? Next year? The market is notoriously impatient. If the recovery doesn't show up in the next earnings season, the stocks will get sold off, regardless of the long-term thesis. This is the classic 'value trap' scenario. You buy a stock because it looks cheap relative to its future earnings, but if those earnings don't materialize on schedule, you're left holding a bag. I've been through enough cycles to know that the market's attention span is short. The AI narrative has been running for over a year now, and it's starting to show signs of fatigue. The 'easy money' has been made. The next phase will require patience and discipline. It will require looking at the fundamentals and being willing to hold through volatility. It will require understanding that the AI trade is not a single trade, but a series of trades across different parts of the value chain. Let me give you a concrete example from my own experience. In 2020, I was heavily involved in yield farming. I was chasing the highest APYs, moving my capital from protocol to protocol, trying to maximize my returns. It was exciting, and for a while, it worked. But then the music stopped. The incentives dried up, and the yields collapsed. I lost a significant portion of my portfolio because I was focused on the short-term yield rather than the long-term sustainability of the protocols. The lesson I learned is that when something seems too good to be true, it usually is. The same principle applies to the AI trade. If a stock is trading at a massive premium to its earnings, you need to ask yourself: is this growth sustainable, or is it a bubble? So, what's my takeaway for the next few months? I think the AI trade is going to be volatile. Nvidia's earnings will be a major catalyst, and the market will react strongly to any news, good or bad. I think the storage and data center trade has merit, but it's not a slam dunk. You need to be selective and focus on companies with strong balance sheets and clear paths to profitability. I also think the rotation into non-AI sectors is a trend worth watching. Banks, miners, and other traditional industries could offer better risk-adjusted returns in the current environment. But most importantly, I think we need to step back and look at the bigger picture. The AI revolution is real, but it's going to take time to play out. The market is always ahead of itself, pricing in the future before it actually happens. This creates opportunities for patient investors who are willing to wait for the fundamentals to catch up with the narrative. It also creates risks for those who are caught up in the hype and fail to see the warning signs. I'll leave you with this thought: the market is a voting machine in the short term and a weighing machine in the long term. Right now, the market is voting for a rotation. It's voting for storage and data centers over semiconductors. It's voting for banks and miners over AI stocks. But the weighing machine is still in the process of determining the true value of these assets. The next few months will be crucial in determining whether the AI trade is a sustainable long-term trend or just another speculative bubble. Keep your eyes on the data, keep your risk management tight, and don't get caught up in the hype. The party might not be over, but the hangover is just beginning.