What is the actual state of crypto and DeFi after the bust — what survived, what's growing, and what was permanent hype
What Actually Survived the Crypto Crash — and What Was Always Going to Disappear
Based on analysis of a 119-node, 419-edge knowledge graph mapping the structure of the cryptocurrency and decentralized finance ecosystem from 2022 to 2026.
The Forest Fire That Cleared the Ground
Imagine a forest where everything grew fast because the soil was temporarily very rich. Then one day in 2022, the soil dried up. Most of what had grown — because it needed that rich soil to survive — died. But a few things had grown real roots. Those survived.
That is roughly what happened to cryptocurrency in 2022. A chain of failures — exchanges collapsing, lending platforms going bankrupt, token prices falling 80-90% — wiped out most of what had been built. The knowledge graph this analysis is based on treats that event not as an ending but as a sorting machine. Everything after 2022 can be understood as an answer to one question: did this thing have real roots, or was it just feeding on temporary richness?
The One Question Everything Gets Sorted By
After the crash, one idea became the test that all surviving projects either passed or failed: does this thing make money from actual activity, or does it only survive by printing new tokens and handing them out?
Think of it like a restaurant. A real restaurant makes money because people pay for meals. A fake restaurant might look profitable for a while if the owner keeps pulling cash from their own savings account to pay the staff — but that only works until the savings run out.
Most of early crypto was the fake restaurant. Projects issued their own coins, paid users in those coins to participate, and called that “yield.” When people stopped believing the coins were worth anything, the whole system unwound.
The projects that survived — and the ones that are growing now — generate fees from real trading activity, real transactions, real demand. A trading platform called Hyperliquid charges fees on every trade. A staking service called Lido takes a cut of Ethereum rewards. A memecoin launchpad called Pump.fun charges fees every time someone creates or trades a new token. None of these depend on giving away their own coins to stay alive.
This sorting principle — real revenue versus made-up revenue — turns out to explain most of what happened to crypto after 2022.
The Accidental Dollar Machine
Here is something the graph shows that was not planned by anyone: all the surviving crypto activity ended up making the US dollar stronger, not weaker.
Stablecoins — digital tokens pegged to the dollar — became the dominant form of money inside crypto. Tether and Circle issue stablecoins that are backed by US Treasury bonds. Every time someone buys a stablecoin, the issuer buys more Treasuries. There are now tens of billions of dollars in Treasuries held this way.
Separately, the US government passed laws making dollar-backed stablecoins easier to use. Other countries competing with the dollar — particularly China, which has its own digital currency — pushed more users toward dollar-denominated stablecoins as an alternative. The geopolitical competition between digital currencies ended up strengthening the dollar rather than displacing it.
No single person designed this. It is what the graph calls an “emergent property” — an outcome that nobody specifically chose but that fell out of many independent decisions all pointing in the same direction. The graph records it as one of the clearest structural findings: crypto, despite its origins as an alternative to government money, has become a mechanism for extending dollar reach.
Ethereum’s Uncomfortable Success
Ethereum is the platform most serious crypto applications are built on. To handle more traffic, Ethereum developers built “Layer 2” systems — think of them as express lanes built alongside a congested highway. The express lanes worked. Traffic increased. Applications scaled.
The problem: the express lanes are so good that most people use them instead of the main highway. The main highway — Ethereum itself — collects tolls based on traffic. When traffic moved to the express lanes, Ethereum’s toll revenue collapsed.
Ethereum had promised its users that the toll revenue would be used to buy back and destroy ETH tokens, making each remaining token more valuable over time (they called this “ultrasound money”). That promise depended on high toll revenue. The scaling solution that made Ethereum more useful also made that promise harder to keep.
The graph records this as an unresolved contradiction. There is no node, no mechanism, no edge in the graph that shows how Ethereum recovers its fee revenue from L2 success. Both the benefit (more utility) and the cost (less revenue for ETH holders) are structural and ongoing.
The Lido Problem: When Infrastructure Becomes a Risk
Lido is a service that lets Ethereum holders participate in network security without locking up their tokens directly. It holds roughly a third of all staked Ethereum — which makes it the single largest participant in Ethereum’s security system.
Ethereum’s security depends on no single entity controlling more than one-third of the network. Lido is approaching that threshold.
The same feature that makes Lido valuable — its scale — is what makes it dangerous. And then it gets more complicated: Lido’s staked tokens can be used as collateral to borrow money, which gets deposited back into the system, which creates demand for more Lido tokens. There is a loop here. The graph’s single highest-weight edge — the strongest connection in the entire 419-edge graph — connects Lido’s dominance to a risk concept called “restaking contagion.”
The graph contains no edges showing anyone stopping this. No governance intervention. No growth cap. The loop runs, and the threshold approaches.
Two Kinds of AI in Crypto — Only One Is Real
There was a wave of tokens issued with “AI” in the name. Most of them went to zero. The graph explicitly tags a confusion: people saw those tokens failing and assumed AI-crypto convergence was hype. But the graph encodes two separate things that were getting mixed up.
The speculative track — tokens that claimed to represent AI value — behaved exactly like the 2017 ICO bubble and the 2021 DeFi token bubble before it. The pattern repeated.
The structural track is different. AI software agents — programs that can act autonomously on the internet — need to pay for things. They cannot easily open bank accounts. Stablecoins and the payment rails built on top of them are a natural fit. The graph encodes this not as hype but as a dependency: before AI agents can make payments at scale, stablecoins need regulatory clarity, and Ethereum’s scaling infrastructure needs to be reliable.
The key distinction is directionality. AI agents are consumers of crypto infrastructure, not crypto-native developments. If stablecoin payments and Layer 2 scaling exist and work, AI agents will use them. Whether they do is a question about AI adoption, not about crypto.
Some Things Nobody Talks About
Bitcoin miners pivoting to AI data centers. When Bitcoin’s mining rewards get cut in half (as they regularly do by design), some miners can no longer afford to run their equipment. But mining facilities have power contracts, cooling systems, and hardware manufacturing relationships. These are also what AI data centers need. The graph encodes a structural connection between Bitcoin’s reward cycles and the physical infrastructure competition for advanced computer chips. The same factories that make Bitcoin mining hardware compete for capacity with the factories making AI chips.
Pump.fun classified alongside serious protocols. A platform designed to let anyone create a throwaway speculative token in seconds is structurally classified in the graph alongside mature financial infrastructure — because it charges fees on every transaction rather than printing tokens to pay users. The graph does not endorse the activity; it records that the fee mechanism is identical in structure to legitimate protocol revenue. What the activity is for is separate from how the protocol captures value.
North Korea’s role in DeFi. A state actor’s theft operations appear in the graph not as an external event but as a node with functional relationships to infrastructure. The methods used to launder stolen funds — using cross-chain bridges, using derivatives platforms — are structural demands on those platforms. The graph records this as a constraint on institutional adoption: when a government’s sanctions office identifies a theft-and-laundering pattern involving specific infrastructure, that infrastructure becomes harder for regulated institutions to touch.
What Remains Unresolved
The graph is honest about what it does not know. Several tensions are recorded without resolution:
Stablecoins increase demand for US Treasury bonds — good for the dollar. But stablecoins also displace bank deposits — bad for the Federal Reserve’s ability to manage the money supply. Both effects are real and ongoing. No equilibrium is encoded.
Hyperliquid is described as the leading example of crypto maturity — a fully on-chain trading system with real revenue. It also shares a structural label with EigenLayer, the primary systemic risk node in the graph, around the question of centralization. The graph records this paradox without resolving whether Hyperliquid’s centralization features are acceptable because of its other properties.
DePIN — projects that use token incentives to build physical infrastructure like wireless networks and data storage — is encoded as competing with major tech company infrastructure spending. But the graph does not record who wins. Competition is asserted; outcome is not.
The Bottom Line
The graph encodes six structural findings that are not obvious from surface-level crypto commentary:
1. The 2022 crash was a filter, not an ending. Everything that survived can be traced to real fee revenue. Everything that did not survive was dependent on token emissions.
2. Crypto accidentally strengthened the dollar. The surviving infrastructure is overwhelmingly dollar-denominated, and stablecoin reserve requirements have created new institutional demand for US Treasuries.
3. Ethereum solved its scaling problem by creating a revenue problem. The L2 success and the fee revenue decline are the same event.
4. Lido is both the most important infrastructure and the most important risk. These are not in tension — they are the same property expressed at different scales.
5. AI-crypto convergence is real but it runs in one direction. AI agents need stablecoins and L2 rails. Whether crypto needs AI agents is a different question.
6. Bitcoin and everything else have structurally separated. Corporate treasuries, sovereign accumulation, ETF flows, and the halving mechanism all point in the same direction: Bitcoin has a different investor base, a different risk profile, and a different narrative than every other crypto asset. This is not a new opinion in the graph — it is the downstream consequence of a dozen independent mechanisms all pointing the same way.