The same week Ethereum's Dencun upgrade cut Layer-2 fees to near zero, one of Wall Street's most respected macro voices dropped a warning that should be reverberating through every crypto trading desk. Abby Joseph Cohen, the Goldman Sachs strategist who called the 1990s bull market, told investors the US economy is "uneven" and that AI investing is "unsustainable."
Most crypto-native analysts brushed this off as legacy finance noise. That's a mistake. Cohen's language is precise. She didn't say the economy is crashing. She said it's imbalanced. And she didn't say AI is worthless. She said the current investment pattern is unsustainable. Those two words โ uneven and unsustainable โ are the exact conditions that precede violent asset repricing.
The Warning in Context
Cohen built her career on data, not narratives. She was one of the first analysts to adopt quantitative valuation frameworks in the 1990s. Her warnings carry weight because she's historically been early but accurate. Her current view centers on a core structural observation: capital is flooding into AI infrastructure while the rest of the economy lags.
This isn't a debate about whether AI will transform industries. It's a liquidity distribution question. Money is being deployed as if AI adoption will be instantaneous and universal. The real world is not cooperating. If you strip away the AI narrative and look at the base economy, you see what Cohen sees: a two-speed system where concentrated sectors boom and everything else idles.
Crypto traders should recognize this pattern instantly. It's identical to what happened in DeFi summer 2020 โ capital flooded into yield farming protocols and the broader market was left with a liquidity vacuum.
The Uneven Economy: A Technical Breakdown
Let's examine what "uneven" means from my surveillance desk. I'm monitoring 7x24, watching capital flows across blockchain networks. The same dynamic Cohen describes in the US macro economy is playing out on-chain.
In Q1 2025, AI-linked token sectors captured disproportionate capital inflows while most alts bled. This isn't a crypto-specific phenomenon. It's the same concentration risk Cohen is flagging in equities. When a few sectors absorb all liquidity, the market becomes vulnerable to a cascading unwind.
During my time monitoring the FTX collapse in 2022, I saw this exact pattern. Capital concentrated in a few venues. When the center failed, everything contracted. The current AI investment cycle is building the same kind of structural fragility.
The Unsustainable AI Investment Pattern
Cohen's second point โ unsustainable AI investing โ deserves forensic scrutiny. "Unsustainable" means current capital deployment cannot be maintained without triggering a correction. It's a mathematical statement, not a value judgment.
Let's break down the fundamentals:
- AI data center buildouts require massive upfront capital with uncertain revenue streams
- The computing resources needed for AI training are doubling faster than infrastructure can deliver
- Corporate AI investments are being funded by debt and equity dilution, not current cash flow
This is the same dynamic I documented during the Arbitrum Nitro migration in 2023. Protocols were spending millions to upgrade infrastructure that had no user demand. The spending was real. The revenue was theoretical. When user growth didn't arrive, token prices repriced downward.
The AI infrastructure narrative is a bigger version of the same problem. The investment is real. The hardware is real. But the actual demand โ businesses paying for AI inference at scale โ remains unproven.
What This Means for Crypto Markets
Crypto assets are uniquely exposed to this macro dynamic. When liquidity contracts in traditional markets, crypto feels the pain first and hardest. Bitcoin's correlation with equities remains high. The current AI capex cycle is supported by the same liquidity conditions that support crypto.
If Cohen is right about the economy and AI investments, the implications for crypto are straightforward. The 42-second liquidity arbitrage windows I captured during the Shanghai upgrade will become multi-day gaps. The risk premium on holding speculative crypto assets will expand as equity markets reprice.
From my monitoring desk, I'm watching for signs of this repricing already. The funding rates on perpetual contracts for AI-related tokens are running hot, the same way they did before the last major correction. The market is priced for a scenario where AI investments pay off smoothly. Cohen's warning suggests the path is much more volatile.
The Contrarian Angle: What Cohen's Warning Reveals
Here's the angle no one is discussing. Cohen's warning may already be priced into traditional markets more than crypto markets. The institutional traders who follow her research have already begun to hedge their AI exposure. Crypto markets, by contrast, have no equivalent risk framework.
This creates a massive information asymmetry. Smart money is quietly de-risking AI-related assets. Crypto markets are still chasing them.
If Cohen is right, we're looking at a scenario where AI infrastructure spending slows down. That won't just hurt AI tokens. It will hurt the entire crypto infrastructure complex because so much of the current narrative depends on AI-agent protocols, decentralized compute networks, and GPU tokenization.
The AI-crypto crossover is the market's biggest narrative, but it's built on top of an investment cycle that a top macro mind just declared unsustainable. If that cycle corrects, the floor collapses for a huge portion of the current crypto thesis.
I've tracked this pattern before. In the Solana outage of February 2023, I identified the congestion was caused by bad actors, not systemic failure. The market narrative was wrong. I'm seeing the same dynamic here. The market narrative is that AI investment is a never-ending growth story. Cohen's data suggests otherwise.
Takeaway: The Signal to Watch
Here's what I'm tracking in the next 60 days. If AI infrastructure companies report earnings with lower-than-expected forward guidance, that's the signal. If NVIDIA's order book shows a slowdown, that's the second signal. If these occur, expect crypto markets to decouple from the AI narrative โ and not in a good way.
The market is priced for an AI utopia. Cohen is telling you the path there has more cracks than the bull case suggests. Based on my experience analyzing the Shanghai upgrade, the FTX collapse, and Solana's outages, I've learned one lesson: when a respected analyst warns about structural fragility, the market always listens later than it should.
Prepare accordingly. Diversify away from AI-narrative tokens. Monitor funding rates and exchange flows. The next 90 days will tell us whether Cohen's warning was prescient or a misread. But the asymmetry of the risk-reward ratio suggests the prudent play is to respect the warning, not dismiss it.