The AI Trade Is Deleveraging: Goldman Sachs Signals the End of the Beta Era

The numbers landed without fanfare, but their implications are structural. Goldman Sachs reports that high-beta momentum portfolios fell 12% in a single week, and their dedicated AI hedge basket dropped 10% over five days. For anyone who has spent the past year watching AI equities trade like a single, monolithic asset class, these figures mark the end of a phase. The crowd is leaving. The question is whether they are leaving a building that is on fire or simply exiting a room that has become too crowded. Based on my experience auditing the 2024 Bitcoin ETF custody structures, the distinction matters. Context is necessary here. Since the narrative shifted in late 2022, the market has treated AI as a unified thesis. Buy the infrastructure, buy the hardware, buy the narrative. It worked. But the investment banks, which are the primary market movers in this cycle, are now signalling a fragmentation. The report identifies a shift in momentum factors: software has replaced semiconductors as the largest weight in the three-month momentum portfolio. Semiconductors and AI complexes have moved into short baskets. This is not a minor position adjustment. It is a reversal of the underlying premise that GPU supply is the only scarce resource. My forensic approach dictates that we reconstruct the ledger. The first entry is the AI hedge basket's five-day, 10% drawdown. The second is the high-beta momentum collapse. The third is the capital rotation. Goldman Sachs explicitly names storage and data centers as tactically the most attractive sectors, because, as they argue, profit recovery is not yet reflected in the share price. This is the core insight. The narrative of training dominance is over. The market is now pricing for the revenue cycle of inference and storage. In my own audit work, I have seen this pattern before. In the 2024 Bitcoin ETF analysis, I standardized a custody risk score because regulatory approval did not equal cryptographic security. Similarly, in this market, a board index rally does not equal profitability. The current divergence between price and earnings in the storage sector is a gap that forensic analysis should treat as an anomaly. If we run the numbers, a profit recovery that is not priced in represents a discrepancy. The discrepancy either resolves upward as the market catches up, or it resolves downward when the fundamentals fail to materialize. The market is betting on the former. My skepticism requires me to note that storage is a cyclical business, and we have not yet isolated the portion of profit recovery driven by AI versus general IT refresh cycles. The margin of error is wider than the market believes. ","A critical factor is the shift in the value chain. The entry is the capital outflow from semiconductors into software. The semiconductor sector is now in the short book. This is a strong signal. It reflects a consensus that the moat of the dominant chipmaker is shrinking, whether through ASIC alternatives or export controls. Consequently, the market is moving from the hardware layer to the application layer. The efficiency gains are now expected in the layer that captures revenue, not just the layer that sells shovels. The contrarian angle is worth examining. The bullish narrative is not entirely wrong. Goldman Sachs is not calling this a bubble. They are calling it an end to the beta phase. That is a distinction with a difference. The previous phase rewarded anyone who simply held a basket. The next phase rewards those who can distinguish between the overvalued and the undervalued. The bulls are right to say the trade is not over. The long-term trajectory of AI infrastructure remains intact. My own audit experience in 2026 with the AI-Agent payment protocol showed me that the underlying technological utility continues to grow, but the market demands proof. The efficiency gains are real, but the identity layer had a flaw. The market is now checking the identity of the revenue claims. The takeaway is one of accountability. The gold, the copper miners, the banks, and the European value stocks are now receiving the overflow capital. That is a warning that the risk appetite for AI infrastructure is saturated at the margin. The short-term catalysts, NVIDIA's Q2 earnings and September conferences, will provide the next signal. If the guidance is weak, the deleveraging will resume. If the guidance is strong, we will see the divergence. The market is no longer pricing the future. It is pricing the present. The numbers have to work. The profits have to appear. For the storage sector, this is the moment of truth. Trust the code, not the press release. Run the numbers, ignore the hype.