The Great Unwind: Goldman's AI Trade Shift and the Quiet Return to Fundamentals
There is a moment in every market cycle when the narrative begins to crack. Not with a crash, but with a whisper. A rebalancing. A quiet rotation that speaks louder than any headline. Goldman Sachs, in its latest strategy note, has offered us such a whisper. It tells us that the AI trade is not dead, but it is changing. And in that change, we find not an ending, but a test of our convictions.
For months, we have watched the AI complex rise as a monolith. Semiconductors, cloud providers, and every company that whispered 'GPU' into their earnings call were rewarded with capital. It was a tide that lifted all boats, a beta trade of unprecedented scale. But the tide, as it always does, has begun to recede. The high-beta momentum portfolio fell 12% in a single week. The AI hedge fund basket dropped 10% in five days. Leverage, that intoxicating fuel of the bull market, is being drained from the system.
This is the context we must understand. We are not witnessing the death of AI. We are witnessing the end of its adolescence. The market is no longer paying for dreams; it is demanding receipts. Goldman's core message is that the era of indiscriminate buying is over, replaced by a more demanding, more discerning phase of stock selection. It is a shift from the collective euphoria of the choir to the solitary discipline of the vigil.
Let us trace the code back to the conscience of this trade. The signals are clear for those who listen to the silence between the blocks. Semiconductors, the high priests of the AI altar, have been placed on the short list. This is not a tactical blip; it is a philosophical repositioning. The market is beginning to question the permanence of the moat around Nvidia's throne. The rise of custom ASICs, the push from AMD, and the geopolitical fog of export controls are no longer theoretical risks. They are being priced in.
Meanwhile, software has taken the crown as the largest weight in the three-month momentum long portfolio. This is a profound shift. It suggests that the value capture is moving from the 'picks and shovels' of the hardware layer to the 'gold miners' of the application layer. The market is betting that AI will finally translate into revenue for software companies, that the promise of AI agents and copilots will materialize into subscription dollars. It is a bet on the 'why' of AI, not just the 'how'.
My own journey through the 2017 ICO audit taught me a painful lesson about the gap between code and conscience. We built systems we believed were trustless, only to find they required more human stewardship than we ever imagined. The same is true here. The market is realizing that the 'trustless' growth of AI infrastructure requires a human-centric check on valuation. It is not enough to build the largest model; you must build a sustainable business.
Goldman's most intriguing signal is the call on storage and data centers. They describe these as the most tactically attractive sectors, with the most significant valuation gaps. The logic is that profit recovery has not yet been fully reflected in the stock prices. This is a direct line to the next phase of the AI lifecycle: inference. We are moving from the training of massive models to the deployment of those models at scale. This requires a different kind of infrastructure. It requires vast storage for model weights and inference caches, and it requires data centers optimized for the latency-sensitive, power-hungry reality of AI inference.
This is where my experience with the MakerDAO governance in 2020 resonates. We fought for transparency in the collateral basket, arguing that a stablecoin should serve as a public good. The same principle applies here. The 'profit recovery' in storage is not just about selling more SSDs; it is about building the foundational layer for a new digital economy. It is about ensuring that the infrastructure serves the human spirit, not just the corporate bottom line.
But here is the contrarian angle, the blind spot in the Goldman narrative. They recommend these sectors because the 'profit recovery is not yet priced in.' But what if that recovery is not as robust as hoped? What if the demand for storage is a cyclical echo of traditional IT spending, not a structural shift from AI? The analysis does not distinguish between AI-driven demand and the broader enterprise refresh cycle. We must be vigilant. Governance is not a vote; it is a vigil. And so is investing. We cannot simply accept the narrative of 'profit recovery' without verifying the underlying data.
The capital rotation to European and Japanese banks, gold miners, and copper stocks is another telling signal. It suggests that the marginal dollar is leaving the crowded AI trade and seeking value elsewhere. This is not a sign of weakness in AI, but a sign of maturity. The easy money has been made; now comes the hard work of fundamental analysis. The copper miners, in particular, are a fascinating play. They are a proxy for the physical reality of AI: the electricity, the cooling, the transmission. It is a reminder that even in the digital age, we build bridges from the ashes of belief, and those bridges are made of physical materials.
As we look to the catalysts ahead—Nvidia's Q2 earnings and the September industry conferences—we must hold space for the digital soul of this market. The earnings report will not just be a number; it will be a confession. It will tell us whether the AI infrastructure build-out is accelerating or plateauing. It will tell us if the 'profit recovery' in storage and data centers is real or a mirage.
We are entering a phase where the protocol must serve the human spirit, not just the speculative appetite. The AI trade is not ending; it is being reborn. It is moving from a speculative fever to a fundamental discipline. This is a healthy, necessary evolution. It is the market's way of asking the hard questions, of demanding that we trace the code back to the conscience.
In this sideways market, the chop is for positioning. The signals from Goldman are not a call to abandon the AI revolution, but a call to be more selective, more thoughtful, more human. We must listen to the community's heartbeat, not just the market's ticker. The opportunity is not in the broad beta, but in the specific alpha that comes from understanding the true value of the infrastructure. Truth is the only immutable asset, and the truth is that AI is moving from the lab to the real world. The question is not if, but who will build the bridges. Let us build them with our eyes open, our values intact, and our focus on the long, patient work of creation.
The unwind is not a retreat; it is a recalibration. It is the market's way of separating the wheat from the chaff, the visionaries from the speculators. As we navigate this period of consolidation, we must remember that decentralization is a practice of radical empathy. It is about understanding the needs of the end-user, the developer, and the investor. The AI trade is entering its most important phase, and it will be defined not by the size of the models, but by the strength of the communities that build and use them. We are not just investors; we are stewards of a new digital frontier. Let us act accordingly.