Goldman Sachs Says the AI Trade Is Over — But the Bleeding Has Just Begun

The numbers hit my screen like a hammer. Goldman's high-beta momentum basket dropped 12% in a single week. Their AI hedge portfolio bled 10% in five days. This is not a correction. This is a deleveraging event. Markets do not care about your sentiment. They care about your position size. When the code bleeds, the ledger keeps the truth. The AI trade that minted millionaires in 2023 is now being unwound by the same machines that built it. I have seen this movie before. It ends with retail holding the bag while smart money repositions. Let me be clear about what is happening. Goldman Sachs, the same institution that was screaming about AI euphoria six months ago, has now placed semiconductors and the AI complex into their short portfolio. Think about that. The trade that everyone piled into is now a crowded exit. The momentum factor, which tracks the last three months of price action, has quietly shifted its largest weight from semiconductors to software. This is not noise. This is a structural signal buried in the order flow. I have been tracking this divergence since my days auditing Solidity contracts in Paris. Back in 2019, I learned that code does not lie. Whitepapers do. The same principle applies to Goldman's positioning data. When a sell-side behemoth starts shuffling its factor weights, you do not argue. You follow the tape. The AI trade is not dead. But the era of buying the whole sector and getting rich is over. We have entered the second phase: the sorting of winners from pretenders. Here is the core insight that most retail traders will miss. Goldman explicitly states that storage and data centers are now the most tactically attractive sectors. Why? Because the profit recovery in these segments has not yet been fully reflected in the stock prices. This is a classic arbitrage setup. The market is still pricing these names based on the 2022 downturn narrative, while the actual earnings are recovering due to AI inference demand. I have seen this pattern before. In the 2020 DeFi Summer, I leveraged my ETH 5x on MakerDAO to mint DAI and deployed it into Compound. The yield was insane, but the volatility kept me awake for weeks. I learned that capital flows to where the earnings are, not where the hype is. The same logic applies here. Let me break down the mechanics. The AI trade has two distinct phases. Phase one was the infrastructure build-out. This was driven by NVIDIA and the hyperscalers throwing billions at GPU clusters. During this phase, the entire sector moved in lockstep. You did not need to pick winners. You just bought the beta. Phase two is the operationalization phase. This is where the actual revenue from AI applications starts to flow. The demand shifts from training compute to inference compute. Inference requires different infrastructure. It requires storage for model weights and KV caches. It requires data centers with high-density racks and efficient cooling. This is where the profit recovery is happening. This is why Goldman is pointing to storage and data centers. Now, let me talk about the elephant in the room: NVIDIA. Goldman lists the Q2 earnings report as a key catalyst. This is a double-edged sword. If NVIDIA beats and raises guidance, the AI complex might get a temporary reprieve. If they disappoint, or worse, guide down, the deleveraging will accelerate. Based on my experience with the Terra collapse in 2022, I know that the market does not care about the past. It only cares about the forward guidance. When Terra fell, I did not panic. I shorted the remaining LUNA positions using options and profited $15,000 as the protocol collapsed. The lesson was simple: the market is a forward-looking discounting machine. NVIDIA's guidance will tell us if the capex cycle is still accelerating or if we have hit peak AI spending. The contrarian angle here is brutal. Everyone is still talking about the AI bubble. They are asking if it will pop. Goldman is not asking that question. They are saying the trade is not over, but the method of extracting alpha has changed. This is a far more dangerous statement. A bubble popping is a clean, fast event. A sector rotation is a slow bleed. It is death by a thousand cuts. The high-beta momentum basket dropping 12% in a week is just the beginning. The real pain comes when the rotation becomes a stampede. Let me give you a concrete example of what I mean. The money flowing out of AI is not disappearing. It is rotating into European and Japanese banks, gold miners, and copper stocks. This is a massive signal. Banks benefit from a steeper yield curve. Gold benefits from uncertainty. Copper benefits from the electrification of everything, including AI data centers. This rotation tells me that the smart money is hedging against a slowdown in AI capex growth while simultaneously positioning for the physical infrastructure build-out that AI requires. This is not a bearish call on AI. It is a bearish call on AI valuations. Now, let me address the storage thesis directly. The memory industry has been through a brutal consolidation. We are left with a triopoly: Samsung, SK Hynix, and Micron. This is a rational oligopoly. They control supply. They control pricing. And now, AI is creating a new demand curve for HBM and enterprise SSDs. The profit recovery in this sector is real. I have been monitoring the HBM supply chain since I started analyzing on-chain options data from Deribit in 2024. The implied volatility versus realized volatility arbitrage opportunities in AI-linked stocks are widening. This is the kind of setup that makes my Quant Institutional Bridge persona salivate. The market is mispricing the persistence of this earnings recovery. The stocks are trading at multiples that assume the 2023 downturn will repeat. The data says otherwise. The data center thesis is similar. The utilization rates are climbing. The rental rates are firming. The power constraints are becoming the new bottleneck. Goldman sees this. The market is still treating these names like utilities with no growth. This is a mispricing. I have seen this exact setup in the crypto mining sector. In early 2021, I led a team of three developers to build a bot for the Bored Ape Yacht Club minting race. We spent $2,000 on RPC nodes to ensure speed, securing 12 NFTs at mint price. I immediately listed them on OpenSea, profiting $40,000 within 48 hours. That was a lesson in infrastructure superiority. The people who control the physical infrastructure control the alpha. Data center operators are the RPC nodes of the AI economy. They will capture the value. Here is the part that most analysts will not tell you. The deleveraging is not over. When Goldman says the high-beta momentum basket dropped 12%, that is just the first wave. The AI hedge portfolio dropping 10% in five days is the second wave. There will be a third wave when the actual earnings reports come in. If NVIDIA's guidance does not blow past expectations, the selling will intensify. The positioning is still too crowded. The retail flow is still too heavy. The options market is still pricing in too much upside. I have seen this exact pattern in the crypto market during the 2022 crash. The deleveraging cascades. The margin calls trigger forced selling. The forced selling triggers more margin calls. It is a vicious cycle. But here is the opportunity. The storage and data center stocks are the first to be sold in the panic, but they will be the first to recover when the dust settles. The profit recovery is real. The balance sheets are improving. The free cash flow is returning. This is not a speculative story. This is a fundamentals story. When the market realizes that these companies are generating actual earnings from AI inference, the multiple will re-rate. The stock prices will catch up to the EPS. This is the trade. This is the arbitrage. Arbitrage is just violence disguised as math. Let me give you my actionable levels. For the storage names, I am watching the relative strength against the broader market. If Micron and SK Hynix can hold their 50-day moving averages while NVIDIA breaks down, that confirms the rotation. For the data center REITs, I am watching the AFFO yields. If the yields compress while the broader market sells off, that is institutional accumulation. I would also keep an eye on the copper miners. They are the ultimate hedge. If AI data center build-out continues, copper demand will surge. If AI capex slows, copper is still supported by the broader electrification theme. The risk here is obvious. If NVIDIA's earnings report is a disaster, everything goes down together. Correlation goes to one. The storage thesis will not save you in the short term. But for a trader with a 3-6 month horizon, the setup is compelling. The market is offering you a discount on assets with improving fundamentals. The key is to size the position correctly and to hedge the downside. I use options for this. I sell covered calls on my long positions to fund puts on the indices. This is the institutional bridge. This is how I survived the Terra collapse. This is how I turned an 80% drawdown into a $15,000 profit. I want to end with a question. The AI trade is rotating, not dying. The question is, are you positioned for the rotation or are you still holding the bag from the beta trade? The market is telling you the answer. The code is showing you the truth. The ledger does not lie. Storage and data centers are the new battleground. The profits are moving there. The smart money is moving there. The question is, will you follow the order flow or will you stay stubborn and watch your P&L bleed out? The choice is yours. I have already made mine. I am long the infrastructure and short the hype. This is the black box. This is the edge.