Title: The Great Rotation: Goldman’s AI Trade Recalibration and the Liquidity Migration No One Is Watching
Article:
History rarely repeats itself, but it often rhymes in the context of market liquidity. Over the past seven days, a specific cohort of AI-focused hedge funds absorbed a 10% drawdown, while high-beta momentum portfolios shed 12%. The immediate instinct is to call this a crash. But my eye is on the horizon, not the hourly candle.
The recent turbulence is not a signal of systemic failure. It is the sound of the market pruning its own excesses. The question is not whether the AI trade is over, but what the capital flow is telling us about the next phase of the cycle.
A recent analysis of a major investment bank’s positioning reveals a distinct structural change in how institutional money is being allocated. The market has moved from a phase of indiscriminate buying of all AI-related assets into a phase of precision-driven, fundamental differentiation.
The key data points are clear. Momentum factors are being recalibrated. Software has displaced semiconductors as the largest weighting in a three-month momentum long portfolio. Conversely, semiconductors and the broader AI complex have now entered the short portfolio. This is not a random event. It is a reflection of the market’s internal ledger, where the perceived risk-reward for chipmakers like Nvidia has shifted relative to software application layers.
Simultaneously, the report flags a tactical recommendation for the storage and data center sector. The thesis is straightforward: valuation gaps are at their most significant, and the profit recovery within these sectors has not yet been fully reflected in share prices. This is classic value-plus-growth logic, applied to the backbone of the AI ecosystem.
The Core: The Disconnect Between Price and Profit
My eye is on the horizon, not the hourly candle. In this context, the horizon belongs to the storage and data center players. From my experience auditing portfolio risk models, I have seen how AI-related capital expenditures can drive infrastructure demand with a lag. The cycle is no longer about the latest GPU; it is about the physical and digital substrate that enables AI to run efficiently.
The logic of the current recommendation is based on a specific disconnect. While the market is focused on the fever pitch of semiconductor sales, the underlying profit recovery for companies like Dell, Supermicro, and Micron is projected to be significant. But the market hasn't priced it in yet. The market is looking at the last three months of volatility, while the valuation metrics are suggesting a different story.
The core insight here is that the "AI trade" is not dead, but it has become an accountant's game.
This is the crucial shift. The phase of multiple expansions, where a company with "AI" in its name would soar, is over. The new phase is about earnings per share, revenue guidance, and free cash flow. The bank’s data suggests that the AI trade is not over, but it is transitioning from a macro-sentiment play to a micro-fundamentals play. The catalyst for this transition is the upcoming NVIDIA earnings report and the industry conferences in September.
The Contrarian Angle: The Non-Trade of the Non-AI World
A particularly interesting narrative in the report is the flow of funds into previously ignored sectors: European and Japanese banks, gold miners, and copper producers. This is a divergence. The natural interpretation is a risk-off rotation, but I see a different layer.
The movement of capital into copper is not a simple "anti-AI" trade. It is a "pro-AI infrastructure" trade. AI data centers require immense physical infrastructure, and that infrastructure requires copper. The market is not fleeing the AI theme; it is decentralizing the theme to find value in overlooked corners of the value chain.
However, the contrarian angle lies in the bank's recommendation itself. The report suggests that the market is looking for "profit recovery" in storage and data centers. But my experience tells me that the "profit" recovery might be a myth. The current AI trade is heavily based on the assumption of a pure efficiency gain. But what if the next stage is a cost, not a profit?
The bust was not an end, but a necessary pruning. In the world of Layer 2s and fragmented liquidity, we learned that scaling often means slicing scarce liquidity into smaller pieces. The same logic applies to AI. As the infrastructure expands, the margins for the hardware providers could be compressed by the very hyperscalers they serve. The profit recovery might not be as robust as the analysts expect.
The Takeaway: Watching the Leverage, Not the Prices
The market is in a sideways/consolidation phase. The chop is for positioning. The key signal to watch is not the price of NVIDIA, but the level of leverage in the AI complex. The recent 10% drawdown in AI hedge portfolios suggests a significant amount of leverage is still being unwound.
The report’s mention of the “leverage” returning from "extremely high levels" implies that the current period is a cooling down, not a reset. The 5-day 10% decline in the AI hedge fund cohort is a wake-up call. If the NVIDIA earnings do not meet the highest expectations, we could see a second round of de-leveraging.
My advice is to watch the code, not the noise. The market is waiting for direction. The signals will come from the print of a balance sheet. The next move is not a trade; it is a positioning for the next cycle. The market is waiting for a direction. The question is whether the direction will come from a copper mine or a data center.
The old cycle is being pruned. The new cycle is being built. The question is not "what is the price of the AI token," but "what is the yield on the data center." That is the trade.