Goldman Sachs' latest trading desk note landed in my inbox at 6:47 AM Melbourne time. The email subject line was innocuous: 'AI Trade: Deleveraging, Not Done.' But buried in the data was a number that stopped me cold. The bank's AI hedge portfolio had dropped 10% in five days. The high-beta momentum basket? Down 12%. I've seen these patterns before—not in equities, but in crypto, where leverage builds silently beneath the surface, then evaporates in a cascade of forced liquidations. The AI trade is now experiencing its first genuine leverage flush since the post-2022 recovery, and the market's response tells us more about institutional behavior than any price chart could.
The numbers are stark. Software has replaced semiconductors as the largest sector in the three-month momentum long portfolio. Semiconductors and AI complexes have moved into the short portfolio. This is a structural shift, not a tactical blip. The same funds that were piling into Nvidia and its cohorts are now rotating into storage, data centers, and even banks in Europe and Japan, gold miners, and copper stocks. The message is clear: the AI trade is entering its value phase, and the market is no longer rewarding narrative without earnings.
This report isn't just about AI equities. It's a map of how institutional capital treats narrative-driven assets when liquidity tightens. And for those of us who have spent years in crypto markets, the parallels are uncomfortable. The same dynamics are playing out in BTC, ETH, and the broader digital asset complex. The ETF-driven inflow narrative is hitting its own version of momentum saturation. We're watching a subtle decoupling from the tech narrative that propped up digital assets for two years.
Goldman's recommendation of storage and data center companies—think Dell, Super Micro, Micron—is built on a simple premise: profit recovery hasn't been fully priced into the stock. This is value investing applied to AI infrastructure. The margin between current valuations and projected earnings is the trade. This is exactly the kind of analysis I wish more crypto investors would apply to Layer 2s and alt-L1s. The market is now asking: What actually generates revenue?
The report identifies a key catalyst: Nvidia's second-quarter earnings and the September industry conference. Goldman is saying, in essence, that the AI trade's future direction is contingent on a single earnings release. That's a fragile market. It's the same fragility we saw in crypto when the BTC ETF approval became the singular narrative driver. The market gets simplified into a single event, and when that event passes, the deleveraging begins.
But here's the part the Goldman note doesn't say explicitly, though it's implied. The recommendation of storage and data centers is a bet on the continuation of AI capital expenditure. It's a bet that data center build-outs continue, that electricity demand skyrockets, that the physical layer of AI expands. And it's the same physical layer that crypto intersects with. When I look at the copper trade, or the storage trade, I'm seeing the market price in the same thing I'm seeing in crypto: a shift from the digital layer to the physical layer.
The contrarian angle is the decoupling thesis. Goldman's note, on its surface, suggests the AI trade is rotating but not dying. But what if the market is telling a different story? What if this rotation is the beginning of a broader sectoral shift, not just a tactical move? The momentum data shows software has replaced semiconductors as the biggest weight in the long portfolio. That's not a rotation. That's a repudiation of the hardware race. The market is saying the chips are the picks-and-shovels, but the gold is in the applications, the data, the software.
Now, look at the crypto parallel. The recent crypto rally has been driven by narratives: BTC as digital gold, ETH as the settlement layer, SOL as the speed demon. But if the market is moving away from the "hardware" of crypto—the infrastructure and layer 1s—and toward the "software"—the applications and the user-facing products—we could see a rotation in the digital asset space that mirrors the equity market. The money that was chasing the next Nvidia in crypto might be moving toward the next Dell. The problem is that most crypto "infrastructure" is overvalued and has no earnings. The trade in AI storage is based on real earnings potential. The crypto equivalent doesn't exist yet.
Based on my audit experience, I've noticed that Layer 2 projects, ZK rollups in particular, are burning cash on proving costs that are unsustainably high. The equivalent of a storage play in crypto would be a decentralized storage project with actual demand and revenue. But the market is still paying for narrative and not profit. When the AI trade in equities rotates because of a earnings gap, the crypto market will face the same rotation, but it won't be towards the equivalent of "storage and data centers" because those projects are either not profitable or are fake. The rotation will just be a crash.
The post-ETF BTC is Wall Street's toy. The peer-to-peer electronic cash vision is dead. The ETF approval has transformed BTC into a macro asset, more correlated with risk than with its original purpose. And the current rotation in AI equities, which is a rotation towards profitability, will eventually hit BTC. If the AI trade is repriced because of earnings, BTC will be repriced because of its liquidity cycle. When M2 contracts or stabilizes, the flow of the ETF slows, and the momentum dies.
The Takeaway: Watch the Flow, Not the Foam
The Goldman note is a mirror. The AI trade is rotating from the froth to the fundamentals. The crypto trade will have to do the same. The market is entering a phase where narrative is a liability and earnings are the only alpha. In the crypto world, that means we need to see protocols with real revenue, not just token emissions. We need to see storage with demand, not just a whitepaper. And we need to see a decoupling from the broader tech narrative, which might be the death of the "digital gold" narrative.
I'm watching the flow of funds from the AI complex to the storage and data centers. I'm watching the copper trade. I'm watching the correlation between Nvidia's earnings and BTC's price. The next quarter is going to be a reckoning. The market is about to find out if the AI trade and the crypto trade are just two sides of the same liquidity coin. If they are, the next 12 months will be a test of which assets have real value. The ones that fail will be the ones that were only trading on emotion.
Emotion is the asset; discipline is the hedge.
