The Great Rotation: Goldman's AI Trade Isn't Dead, It's Just Learning to Read

The signal arrived on a Tuesday, buried in a Goldman Sachs note that most retail traders will never read. The AI hedge basket had just bled 10% in five days. The high-beta momentum portfolio, the one that had been printing money since ChatGPT went mainstream, was down 12%. On the surface, this looks like capitulation. The end of the AI trade. But Goldman's analysts are saying something else entirely: the AI trade isn't over. It's just learning to read. I've been tracking this narrative shift from my desk in Seoul, watching the data streams collide. The static is deafening. But if you filter it right, you can hear the signal. And the signal here is not about death. It's about rotation. The market is not abandoning artificial intelligence. It's abandoning the lazy version of it. Let's rewind the tape. For the past eighteen months, the playbook was simple: buy anything with a GPU in its supply chain, hold, and watch your portfolio mimic a hockey stick. The narrative was a monolith. Nvidia was the sun, and every other AI stock was a planet reflecting its light. But momentum, that fickle beast, has started to rebalance. Goldman's data shows that software has now overtaken semiconductors as the largest weight in the three-month momentum long basket. Meanwhile, semiconductors and the broader AI complex have been shoved into the short basket. The market is effectively saying: we've priced in the chips, now show us the applications. This is where my "signal-in-noise" methodology kicks in. The raw data is clear, but the interpretation requires a deeper dive. Goldman is pointing to a specific pocket of opportunity: storage and data centers. Their logic is simple and, frankly, beautiful in its clarity. The profit recovery in these sectors hasn't been fully reflected in the stock prices. There's a valuation gap. The market has been so obsessed with the compute layer that it forgot the memory layer. But AI models don't just need to think; they need to remember. Every training run, every inference request, every RAG pipeline—it all requires massive, scalable storage. The narrative is shifting from "who builds the brain" to "who builds the memory." Based on my experience covering the infrastructure wars, this feels like a classic second-order play. In 2021, everyone was mining Ethereum. But the real money was made by the people selling the shovels—the GPU distributors, the cooling system manufacturers, the rack suppliers. We're seeing the same pattern now. The hyperscalers are buying GPUs by the truckload, but they're also buying storage arrays, networking gear, and data center capacity. The profit recovery in these areas is real, but the market's attention has been monopolized by the chip narrative. But here's the contrarian angle that keeps me up at night. Goldman's recommendation is based on a "profit recovery" thesis. They believe earnings will catch up to the infrastructure buildout. But what if the buildout slows? The key catalyst they identify is Nvidia's Q2 earnings and the September industry conferences. If Jensen Huang walks on stage and delivers a cautious outlook—if he even hints at a digestion period for AI capex—the entire storage and data center trade gets repriced in a heartbeat. The "profit recovery" narrative is a promise, not a fact. And in this market, promises are fragile. There's another layer to this rotation that most analysts are missing. Goldman notes that capital is flowing to "previously overlooked areas"—European and Japanese banks, gold miners, and copper stocks. This is the tell. Copper is the metal of electrification. AI data centers are power-hungry beasts, and that power needs to travel through copper. The fact that copper miners are catching a bid suggests that the market is starting to price in the physical reality of AI infrastructure. It's not just about silicon anymore. It's about energy, cooling, and raw materials. The AI trade is becoming a commodities trade. This is where my "Trust, but Verify" instinct kicks in. I've spent years auditing custody solutions and supply chains. The AI narrative is no different. We need to verify that the "profit recovery" in storage is actually tied to AI demand, not just a cyclical bounce. Is it HBM (high-bandwidth memory) that's driving the recovery, or is it traditional NAND flash? The answer matters. If it's HBM, then the trade is directly tied to Nvidia's roadmap. If it's NAND, then it's a broader cyclical story that could be more resilient. The risk matrix here is stark. The top risk is a Nvidia earnings miss triggering a second wave of deleveraging that drags down the storage and data center names. The probability is medium, but the impact is high. The second risk is that the "profit recovery" thesis gets falsified. Companies in the storage sector might guide lower, citing inventory digestion or pricing pressure. The third risk is macro: a liquidity squeeze or geopolitical shock that forces a broad de-risking. But the opportunity set is equally compelling. The first opportunity is to build a position in storage and data center leaders—names like Micron, Dell, and Super Micro—ahead of the earnings catalyst. The second is to use the Nvidia earnings volatility to your advantage, perhaps through options strategies that benefit from a large move in either direction. The third, and perhaps most interesting, is to follow the capital flow into copper and other AI-adjacent resource plays. This is a longer-duration bet, but it aligns with the physical buildout of the AI economy. The signals to track are clear. In the short term, watch Nvidia's Q2 earnings on August 28th. The guidance will set the tone for the entire complex. In September, watch the industry conferences for any hints of a technology roadmap shift. In the medium term, monitor the EPS revisions for the storage and data center sectors. If Goldman is right, we should see upward revisions over the next three to six months. And in the long term, watch the flow of capital into non-AI sectors. If the rotation is permanent, it signals that the AI trade has matured from a speculative mania into a fundamental growth story. I'll be honest with you. I'm cautiously optimistic. The deleveraging we're seeing is healthy. It's the market purging the weak hands and the leveraged tourists. The narrative is evolving from "AI will change the world" to "AI is changing the world, and here's who's getting paid for it." That's a more mature, more sustainable narrative. But it's also a more demanding one. It requires fundamental analysis, not just momentum chasing. It requires reading the footnotes, not just the headlines. The AI trade isn't dead. It's just learning to read. The question is whether you're ready to read along with it. The next chapter is being written right now, in the data centers and the storage arrays, in the copper mines and the power grids. The signal is there, buried in the static. Are you listening? `,