The party isn't over, but the DJ has definitely changed the track. That's the message from Goldman Sachs' latest deep dive into the AI trade, a report that reads less like a eulogy for the sector and more like a tactical reshuffling of the deck. While the overarching narrative of AI-driven growth remains intact, the easy money phase—where simply owning any AI-adjacent stock guaranteed returns—has officially ended. The streets are whispering that the new alpha is hiding in the plumbing, not the processors. We didn't dodge the chaos; we danced through it.
Goldman's report is a cold splash of water for anyone still nursing the highs of the last year. The data shows a dramatic deleveraging event: the AI hedge basket was down 10% in just five days, and the high-beta momentum basket took an even harder 12% hit. This is the sound of leverage exiting the room, quickly. However, the key phrase in the report is not "sell everything," but rather that the era of achieving outsized returns simply by buying the broad AI sector is changing. This isn't a rejection of the technology; it's a rejection of the lazy, beta-driven approach to trading it.
So, where is the smart money moving? Goldman's momentum factors are painting a picture that would have been unthinkable just a quarter ago. Software has dethroned semiconductors as the top weight in the three-month momentum long basket. Meanwhile, the once-dominant semiconductor and AI complex have actually flipped into the short basket. That is a massive tectonic shift. It signals that the market is now betting on the application layer and the tools that enable AI to be used, rather than just the raw chips that power it. This isn't just a rotation; it's a maturation signal.
The most actionable insight from the report is a clear tactical recommendation: storage and data centers. The logic here is beautifully simple. Goldman argues that these sectors exhibit the most significant valuation gap, where the profit recovery has not yet been fully reflected in the stock prices. This is a value-plus-growth play, seeking companies that are absolutely critical to the AI build-out but have not yet been bid up to nosebleed valuations like some of their pure-play peers. The party is moving to the back room, where the infrastructure lives.
The narrative is clear. The AI trade is shifting from a story about raw compute to a story about the physical and digital infrastructure that makes that compute useful. It's about the data centers that house the servers and the storage solutions that hold the vast oceans of data needed for training and inference. This is a bet on the maintenance and expansion of the AI world, a world that requires massive capital expenditure. It’s a mature move, a sign the industry is growing up and the markets are getting smarter about where the real value lies.
But where does the risk hide in this new narrative? The contrarian angle here is that this move into "picks and shovels" is also a form of risk-off within the AI theme itself. By favoring storage over the high-flying semiconductor complex, the market is expressing a preference for more tangible, cash-flow-generative assets. It is a defensive move dressed up as a tactical opportunity. The survival of the AI trade isn't a question of innovation anymore; it's a question of financial engineering. This is a high-stakes game where the market is asking for profits, not promises.
Furthermore, the report's observation that funds are flowing into non-AI sectors—European and Japanese banks, gold miners, and copper stocks—is a critical signal. This suggests that while the AI trade is on pause, capital isn't waiting idly. It's seeking alternative sources of alpha. This could be a simple risk-off move, but the mention of copper is too specific to ignore. Copper is the literal conduit for the electrification of AI data centers. This suggests the market is beginning to price in the secondary, tangible effects of AI infrastructure, moving beyond the digital and into the physical.
Let's talk about the elephant in the room: Nvidia. Goldman Sachs identifies Nvidia's Q2 earnings report and the September industry conferences as the key catalysts that will define the next move in AI. This is the ultimate stress test. If Nvidia delivers a blowout quarter and, more importantly, gives strong guidance on future AI capital expenditures, the risk appetite could flood back into the whole sector. But if they stumble, or the guidance is weak, that second wave of deleveraging could be brutal. The market is a powder keg, and Nvidia is holding the match.
The risk matrix is top of mind for anyone in the trenches. The primary threat is that a weak Nvidia report triggers a second wave of de-leveraging that drags down the storage and data centers that Goldman just recommended. This is the core tension of the market: a rising tide lifts all boats, but a falling tide can leave even the best-placed ships on the mud. The second risk is that the "profit recovery" thesis for storage is a narrative that the fundamentals don't support. The third is the macro environment—liquidity tightening or geopolitical shocks that could force a broader exodus from risk assets, AI included.
Despite these risks, the opportunities are clear. The path forward is not about abandoning the AI thesis but about being a smarter participant. The strategy is not to buy the whole basket but to cherry-pick the laggards with strong fundamentals and a clear path to earnings. The action is to be tactical, to use the volatility to position for the long-term infrastructure build. The play is to look for names that can survive and thrive, even if the broader index corrects. This is a market for selectors, not just investors.
The opportunity set is about identifying those specific players. The low-hanging fruit is in the storage and data center names that are not yet fully repriced. We are looking at companies with stable revenues, real products, and a direct line to the AI data center build-out. The second is a tactical play around the Nvidia earnings. The third is a longer-term macro bet on the AI-induced demand for physical resources, like copper and energy. These are the "AI plus" trades that the market is just beginning to discover.
Walls crumble when the party truly begins. The AI party isn't over; it's just moved venues. The past few months have been a brutal flush, a test of conviction. The reports from the likes of Goldman Sachs are not a death knell but a new map. The market is now trading a more mature narrative, one where value is derived from the infrastructure that supports the digital revolution, not just the hype of the revolution itself. The questions now are about which names have the grit to survive the volatility. The guest list was wrong; the vibe is right. The network breathes in Prague, but it computes in the data centers. The shift is from theory to practice, and it’s a dance we all have to learn.