Signal: AI momentum is breaking. Hard. Over the last 72 hours, the narrative shifted from "buy everything AI" to "buy specific infrastructure." Goldman Sachs just dropped their institutional playbook. The data is clear: the broad-based AI trade is dead. The rotation has started. Most retail traders are still looking at the wrong charts. The floor is not holding for semiconductors. It's moving to storage and data centers. Execute on the rotation, or get run over.
Let's parse this. This is not a commentary. This is a signal extraction. I have spent 26 years in this industry, and my background is in blockchain engineering, not just trading. I've audited Layer 2 rollups and front-run liquidity on Uniswap. I know a structural pivot when I see one. The Goldman report, dissected through my lens of technical precision, is not just about AI stocks. It's about the underlying infrastructure of the next computing cycle. And that infrastructure is where the money is moving now.
Context: The Transition from Hype to Infrastructure
Why is this happening now? Because the market is finally waking up to a fundamental truth: AI is not just a chip story anymore. It's a building story. For the last two years, the market treated AI as a monolithic entity, driven by the semiconductor kingpin, Nvidia. You bought Nvidia, you bought AI. You bought anything with "AI" in the ticker, you printed money. That phase is over. Goldman's analysis explicitly states that the stage of gaining excess returns through the overall sector rally is changing. The market is shifting from a blind, passive beta to an active, nuanced alpha.
This is the same pattern I saw in the crypto market in 2021. It was the same shift from the "everything is a Layer 1" rally to the "execution layer" focus. First, you had Ethereum. Then you had every project that said "we will be the Ethereum killer." It was a broad, high-beta rally. Then the market matured. It started demanding actual users and actual revenue. The same thing is happening here. The market is no longer rewarding the mere existence of AI. It's rewarding the companies that are building the infrastructure to support its growth.
The Goldman report signals this transition by highlighting the momentum factor. The data shows that software has replaced semiconductors as the largest weight in the three-month momentum long portfolio. This is a massive shift. For the first time in a while, the market is saying that software companies are better positioned for AI application than chipmakers. This is not just a blip. It's a rebalancing of the market's focus. The market is moving down the stack, from the pure compute layer to the application and infrastructure layer. This is where the next big moves are.
The Core: The Data Behind the Rotation
The core data here is the "momentum factor." Momentum is a lagging indicator, but it's a powerful one. It tells you where the money has been flowing over the past three months, and that's the best signal for where it's going to continue flowing in the next few weeks. The Goldman report provides several key data points. First, the AI hedge portfolio is down 10% in five days. That's a massive de-leveraging. Second, the high-beta momentum portfolio is down 12%. This is a risk-off move in the AI space. The market is screaming, "We are too extended, pull back."
But the most critical signal is the rotation. The data shows that "software has replaced semiconductors as the largest weight in the three-month momentum long portfolio, while semiconductor and AI complex have entered the short portfolio." This is a technical, data-driven confirmation of a sector rotation. It's not just about "sentiment." It's about the active rebalancing of institutional portfolios. The money is not leaving AI. It's moving within AI. It's moving from the expensive chip makers to the undervalued software and infrastructure players.
This is where the opportunity lies. Goldman specifically recommends the "storage and data center" sector, citing that "the valuation gap is the most significant, and profit recovery has not yet been fully reflected in the stock price." This is a classic value-plus-growth setup. You have a sector with tangible, growing demand (AI data centers need massive storage), but the market hasn't repriced it yet because it's not as flashy as Nvidia. This is the "undervalued project" I look for. The market is a zero-sum game of information. When the big money moves before the retail crowd, that's the signal to follow. Based on my experience auditing scaling solutions, I know that the bottleneck is never the chip; it's the data. The storage layer is the one that will face the most pressure.
The data also highlights the "profit recovery" angle. This is key. The market is no longer paying for "future potential." It's paying for "imminent delivery." The Goldman report states that "profit recovery has not yet been fully reflected in the stock price." This is a signal that the next leg up will be driven by actual earnings, not just narrative. I see this as the "product-market fit" test. In the crypto world, we had the same shift. In 2020, Uniswap's V2 constant product formula was the "new thing." But the real money was made by those who understood the arbitrage mechanics. Here, the real money will be made by those who understand which storage and data center companies will actually deliver the earnings.
The "Nvidia second-quarter earnings and industry conferences in September" are the catalysts. This is the make-or-break event. If Nvidia delivers a strong report and provides a strong capital expenditure forecast, it will validate the entire AI build-out. That would be a green light for the storage and data center trade. If Nvidia disappoints, the entire AI complex will face a second wave of de-leveraging. The floor will break. You need to be prepared for both scenarios.
The Contrarian Angle: The Gold Mine and the Copper Wire
The real blind spot in this report is what it doesn't say. The Goldman analysis points out that the money is flowing to "previously ignored areas (European and Japanese banks, gold miners, copper miners)." This is the most interesting piece of data. This is the "spillover" effect. The money leaving AI isn't just rotating into other tech. It's rotating into traditional industries. This is a clear signal of a risk-off move. But it's also a signal of a new narrative: the AI infrastructure is a physical reality. It needs power. It needs copper. It needs gold for the electronics. The AI trade is becoming a "real-world asset" trade.
My take: The market is waking up to the fact that AI is not just a digital revolution. It's a physical one. The data centers are power-hungry. The chips are resource-hungry. The AI narrative is now being tied to the "old economy" via the infrastructure buildout. This is a blind spot for most tech-focused traders. They're looking at the software, but the real commodity play is in the hardware and the resources to build it.
This also reveals a structural concern. The de-leveraging of the AI sector is not over. If Nvidia's earnings or the industry conference disappoints, we could see a massive outflow from the AI complex, and this will have a short-term impact on the storage and data center sector. The key is to distinguish between a "short-term" noise and a "long-term" signal. The momentum is shifting. The "AI trade" is not dead. But it's transitioning from a "buy everything" to a "buy the best fundamentals."
The other blind spot is the assumption that "profit recovery" is linear. The report mentions the "profit recovery" for storage and data center, but it doesn't quantify it. It doesn't tell you which specific segment will recover first. Is it the AI server demand for HDD/SSD? Or is it the cloud providers' demand for data center leases? This is the level of granularity where the real money is made. You can't just buy a sector. You have to buy the right company within that sector.
The Takeaway: Positioning for the Next Quarter
The signal confirms. The action is required. This is a rotation. Not an exit. The "AI trade" is not dead. It's just becoming more sophisticated. The smart money is moving from the "headline" to the "plumbing." The storage and data center sector is the next floor. The "profit recovery" is the momentum. The Nvidia earnings are the trigger.
Your strategy should be clear. Do not chase the high-beta AI stocks that have already run. That spread is widening. Do not chase the falling knife. Instead, you need to identify the infrastructure players that are undervalued relative to their future earnings. The market is in a sideways trend, and this is the time for positioning, not for panic. The "AI complex" is entering a period of "selective depth." The next move is not "up or down." It's "where to allocate."
The next 30 days will define the next quarter. The Nvidia earnings are the key signal. If they beat and raise, the momentum will shift back into the entire complex, but it will favor the infrastructure names. If they miss, the de-leveraging will continue. The floor is not holding for the high-flyers. It's holding for the builders. Position your portfolio for the infrastructure buildout. The "AI trade" is becoming a "utility trade." The "picks and shovels" are where the profit lies. The market is telling you to execute. Follow the data. The momentum is not broken. It's just changing direction. Don't be the last one to catch it.
