Goldman Says the AI Trade Isn't Dead. It's Just Getting Harder.

The AI hedge basket at Goldman Sachs fell 10% in five days. The high-beta momentum portfolio dropped 12% in a single week. Leverage in the AI complex is unwinding from extreme highs. This is not a crash. This is a deleveraging event. And it changes everything about how you should be positioned. I have seen this movie before. In May 2022, when Terra collapsed, the market didn't just lose a stablecoin. It lost a narrative. The same thing is happening now, but the narrative isn't dead. It's maturing. Goldman's latest note makes this clear: the AI trade is not over, but the phase where you buy the whole sector and get rich is done. The beta is gone. The alpha is all that remains. Let me break down what Goldman is actually saying, because the headline misses the nuance. The bank's strategists are telling clients that the AI trade has entered a new phase. The first phase, which ran from 2023 through the first half of 2024, was characterized by a rising tide lifting all boats. Liquidity was loose. The AI narrative was fresh. You could buy any semiconductor name, any AI-adjacent stock, and watch it appreciate. That phase is over. What replaces it? Stock selection. Fundamental differentiation. The market is no longer paying a blind premium for the AI vision. It is demanding AI revenue. This is the transition from narrative-driven investing to fundamentals-driven investing. And it is brutal for those who don't adapt. The most striking signal in Goldman's note is the positioning shift. Semiconductors and the AI complex have moved into the short portfolio. Let that sink in. The most beloved trade of the past eighteen months is now a short. Meanwhile, software has replaced semiconductors as the largest weight in the three-month momentum long portfolio. The rotation is not subtle. It is a structural repositioning. What does this tell me? The market is pricing in a slowdown in AI training demand. The GPU scarcity narrative is cracking. Custom ASICs are eating into Nvidia's monopoly. Export controls are shrinking the addressable market. The days of unlimited demand for H100s are numbered. I have been saying this since I audited smart contracts in 2017 and learned that hype always outruns fundamentals. The question is not whether Nvidia's growth slows. It is when the market starts pricing that slowdown. That moment appears to be now. But here is the contrarian angle that most retail traders will miss. Goldman is not bearish on AI. They are bearish on the old way of trading AI. The bank explicitly identifies storage and data centers as the most tactically attractive sectors. Why? Because profit recovery has not yet been fully reflected in stock prices. This is a direct call on the AI value chain shifting from compute to storage and inference infrastructure. Think about what this means technically. AI training requires massive compute. But AI inference requires massive storage. Model weights, training data, KV caches, inference clusters. The storage requirements for AI inference are an order of magnitude different from traditional workloads. And the market has not priced this in. Goldman sees it. I see it. The question is whether you see it. My own experience in the 2020 DeFi leverage play taught me this lesson the hard way. I deployed $50,000 into yield farming strategies on Compound and Uniswap. I rebalanced every four hours. I got liquidated for $12,000 when oracle manipulation hit. The pain of that loss taught me that on-chain mechanics behave differently than paper models. The same principle applies here. The market's paper model of AI infrastructure spending is wrong. The real mechanics of AI deployment favor storage and data center operators in ways the market has not yet priced. Goldman's note also reveals capital rotation into overlooked areas: European and Japanese banks, gold miners, copper miners. This is not random. Copper is the transmission metal for AI data centers. Gold is the hedge against AI-driven inflation. Banks are the beneficiaries of a steeper yield curve. The smart money is not leaving the AI trade. It is hedging it. It is finding value in the second-order effects of AI infrastructure buildout. Here is what I don't see in the retail commentary. Nobody is talking about the storage oligopoly. Samsung, SK Hynix, Micron. Three players control the HBM market. HBM is the memory stack that sits next to every AI accelerator. The supply is concentrated. The demand is exploding. And the stocks are trading at valuations that do not reflect the profit recovery that is already underway. This is the kind of structural edge that I look for. It is not a narrative. It is a supply-demand imbalance that is verifiable in the financial statements. Now, let me address the elephant in the room. Nvidia's Q2 earnings, due at the end of August, are the next catalyst. Goldman lists this as a catalyst, not a risk event. That tells me they expect positive signals. But I have been in this game long enough to know that expectations are already high. The bar is not just beating estimates. It is raising guidance. If Nvidia guides below whisper numbers, the deleveraging accelerates. If they guide above, we get a relief rally. Either way, the volatility will be extreme. Position accordingly. The September industry conferences are the second catalyst. These events will provide color on the AI supply chain that earnings calls cannot. I will be watching for commentary on HBM pricing, data center utilization rates, and inference demand. These are the leading indicators that matter. Not the narrative. Not the hype. The numbers. Let me be direct about the risks. The deleveraging may not be over. The high-beta momentum portfolio fell 12% in a week. That is a violent move. It suggests forced selling. When leverage unwinds, it tends to overshoot. If Nvidia disappoints, we could see another leg down. The storage and data center trade could also fail if the profit recovery proves weaker than expected. I have been burned before by trusting sell-side narratives. I will not do it again without verification. My approach is simple. I track the weekly momentum factor changes. I monitor the AI hedge basket performance. I watch the storage names for earnings confirmation. I do not trade narratives. I trade data. The market doesn't care about your opinion. It cares about your position sizing and your risk management. Here is my takeaway. The AI trade is not dead. It is evolving. The beta phase is over. The alpha phase has begun. Storage and data centers are the highest-conviction tactical plays. Software is the momentum leader. Semiconductors are the short. Nvidia's earnings are the catalyst. The market is telling you something. The question is whether you are listening. I don't chase narratives. I chase liquidity flows. And right now, the liquidity is flowing from compute to storage, from hardware to software, from the crowded trade to the overlooked one. Position accordingly. Or get left behind.