The numbers hit the screen at 14:32 on a Tuesday. The AI hedge fund basket had shed 10% in five days. The high-beta momentum basket was down 12%. For anyone who had been watching the capital flows into AI-related equities since late 2023, this was not a blip. This was a structural shift in the ledger of market sentiment.
Goldman Sachs calls it a 'deleveraging.' I call it a repricing event. The difference matters because one implies a temporary correction, while the other suggests a permanent change in how the market values the AI narrative. Hype is a mask; the ledger is the face beneath it. And the ledger is showing that the era of indiscriminate AI buying is over.
This is not a story about technology. It is a story about capital allocation, about the cold mechanics of momentum factors, and about what happens when a market narrative matures faster than the underlying earnings. As someone who has spent years tracing the flow of funds across blockchain networks, I recognize the pattern. It is the same pattern I saw in the 2017 ICO mania, the same pattern in the 2021 NFT bubble. The only difference is the asset class.
Context: The End of the Broad Beta Trade
The AI trade, as it has existed for the past eighteen months, was a simple bet. Buy the basket. Ride the wave. The thesis was that artificial intelligence would transform the global economy, and that every company with a GPU or a data center would benefit. It was a macro trade, not a micro one. You didn't need to pick winners. You just needed to be long the sector.
Goldman's latest report, which I have dissected line by line, confirms that this phase is ending. The bank's momentum factors show a clear rotation. Software has replaced semiconductors as the largest weight in the three-month momentum long basket. Semiconductors and the 'AI complex' have moved into the short basket. This is not a minor adjustment. It is a fundamental reallocation of risk.
The trigger for this shift is not a single event but a confluence of factors. The first is valuation. After a historic run, the price-to-earnings ratios of many AI names have become detached from their actual earnings. The second is the law of large numbers. It is mathematically harder for a company like Nvidia to grow at 50% annually when its revenue base is already in the hundreds of billions. The third is a subtle change in investor psychology. The market is no longer asking 'what can AI do?' It is asking 'what has AI done for my portfolio lately?'
Goldman's recommendation is to pivot towards storage and data centers. The bank argues that these sectors have the most significant valuation gap, with profit recovery not yet fully reflected in stock prices. This is a classic value-plus-catalyst play. It is also a recognition that the AI buildout is not over. It is just entering a new phase.
Core: A Systematic Teardown of the Rotation
Let me be precise about what the data shows. The Goldman report is built on two primary pillars: momentum factors and fund flows. Both tell the same story, but they tell it in different languages.
The momentum data is the most revealing. In the three-month window, software has overtaken semiconductors as the dominant long position. This is a significant shift. Semiconductors, particularly the AI chip makers, have been the darlings of the bull market. Their move to the short basket suggests that the market believes their run is over, at least for now. The question is whether this is a cyclical downturn or a structural change.
My analysis leans towards the latter. The semiconductor trade was predicated on a supply shortage. The narrative was that AI demand would outstrip chip supply for years, creating a pricing power bonanza for manufacturers. That narrative is now being tested. The supply chain is catching up. New fabrication plants are coming online. And there is a growing concern that the hyperscalers—the companies building the massive data centers—are reaching a point of diminishing returns on their capital expenditures.
This is where the storage and data center recommendation comes in. Goldman is not just picking a new sector. It is identifying where the next wave of AI capital expenditure will land. The logic is simple. The chips are already in the ground. Now you need the infrastructure to support them. You need the storage to hold the training data. You need the data centers to house the servers. You need the power infrastructure to keep it all running.
I have seen this pattern before. In the blockchain world, we called it the 'pick and shovel' trade. During the gold rush, the people who made the most money were not the miners. They were the ones selling the picks, the shovels, and the blue jeans. The same principle applies here. The AI gold rush is still on, but the easy money in the miners (the chip makers) has been made. The next phase belongs to the infrastructure providers.
However, I must apply my own forensic skepticism to this thesis. The 'profit recovery' that Goldman cites is an expectation, not a reality. The bank is betting that the earnings of storage and data center companies will be revised upwards over the next few quarters. This is a plausible bet, but it is not a certainty. The AI capital expenditure cycle is notoriously lumpy. A single disappointing earnings report from a major hyperscaler could delay projects and push the profit recovery further out.
Every transaction leaves a scar on the chain. The same is true for capital flows. The scars on the momentum charts are clear. The question is whether they will heal or become permanent.
The Contrarian Angle: What the Bulls Got Right
It would be easy to read this report as a bearish signal for AI. That would be a mistake. Goldman is not saying the AI trade is over. It is saying the AI trade is evolving. The bank explicitly states that the 'AI trade is not over.' This is a crucial distinction.
The bulls were right about the transformative potential of AI. They were right about the massive capital expenditures. They were right about the secular growth trend. Where they were wrong was in assuming that the entire sector would move in lockstep forever. Markets do not work that way. They are dynamic systems that constantly reprice risk.
The rotation towards software is particularly telling. It suggests that the market is beginning to differentiate between the companies that build the AI infrastructure and the companies that use it to generate revenue. The software companies are closer to the end user. They are the ones monetizing the AI capabilities. If AI is truly transformative, the value will eventually accrue to the application layer, not just the infrastructure layer.
This is a lesson I learned during the FTX collapse. When I traced the on-chain movements of $1.8 billion in misappropriated funds, I saw that the market had been pricing in a narrative of invincibility. The reality was far more fragile. The same principle applies here. The market has been pricing in a narrative of infinite AI growth. The reality is that growth will be uneven, and the winners will be those who can execute.
Numbers have no emotions, only consequences. The consequence of the current rotation is that investors need to be more selective. The days of buying the basket are over. The days of picking the right names have begun.
Takeaway: The Accountability Call
The AI trade is not dead. It is being repriced. The market is moving from a phase of broad-based speculation to a phase of fundamental differentiation. This is a healthy correction, not a terminal decline.
The key catalyst to watch is Nvidia's second-quarter earnings report and the industry conferences in September. These events will provide a reality check on the AI capital expenditure cycle. If Nvidia delivers strong guidance, the momentum could shift back. If it disappoints, the deleveraging could accelerate.
My advice is to follow the capital flows. The money is moving from the chip makers to the infrastructure providers. It is moving from the US tech giants to European and Japanese banks, to gold miners, and to copper stocks. This is not a retreat from AI. It is a diversification of the AI trade.
The blockchain is never silent. Neither is the market. The signals are there for those who know how to read them. The question is not whether AI will change the world. It is whether you are positioned to profit from the change. The ledger will tell you the answer. You just have to be willing to look.