This matters for concentration risk in a second-order way. When a category of assets — whether ICO tokens in 2017 or AI tokens today — becomes associated with a particular exchange or a particular liquidity pool, the failure of that venue does not merely destroy the exchange. It destroys the price signal for an entire asset category, triggering forced selling across correlated positions on other platforms. The contagion mechanism is not linear; it is reflexive.
"A token grasp of the AI boom" — Bloomberg's framing — captures the essential problem: investors are acquiring exposure to a complex technological trend through instruments whose market structure is fragile, whose issuer disclosures are often minimal, and whose trading infrastructure is concentrated in ways that amplify rather than absorb shocks.
The 2020–2022 Template
Students of The Great Crypto Bubble (2020-2022) will recognise the pattern. During that cycle, the expansion phase was characterised by an explosion of new asset categories — DeFi tokens, NFTs, algorithmic stablecoins — each of which traded heavily on a small number of platforms. Binance, FTX, and Coinbase collectively dominated global volume. When FTX collapsed, the contagion was not limited to FTX customers. It propagated through every asset that had significant liquidity on FTX's order books, through every fund that had counterparty exposure to Alameda Research, and through every retail investor who had treated the exchange's native token, FTT, as a store of value.
The parallel to today is not that any specific exchange is about to fail. It is that the structural conditions enabling such a failure — concentrated volume, opaque counterparty relationships, and a new cohort of speculative assets with thin liquidity — have not been structurally resolved. Regulatory frameworks in the United States and European Union have advanced, but they have not yet produced the kind of exchange-level transparency, segregated custody requirements, or real-time proof-of-reserves standards that would make concentration risk genuinely manageable. For crypto market cycle data tracking volume distribution across venues, the picture remains one of persistent oligopoly.
The dynamic is self-reinforcing. Macro stress drives investors toward liquidity, which concentrates selling pressure on the largest and most accessible venues, which amplifies price moves in thinly traded tokens, which triggers margin calls and forced selling across correlated positions. This is the cascade sequence that regulators have identified but not yet structurally prevented. For a broader view of current bubble risk indicators across asset classes, the crypto concentration signal sits alongside several other elevated readings.
What Would Change the Calculus
Three structural developments would materially reduce exchange-concentration risk: mandatory real-time proof-of-reserves with independent attestation; enforceable segregation of customer assets from exchange proprietary trading; and deep, liquid decentralised settlement infrastructure capable of absorbing volume during centralised exchange stress events. None of these is fully in place across the major jurisdictions. Decentralised exchanges have grown in sophistication since 2022, but they still account for a minority of global volume and face their own smart-contract and oracle risks.
Until these structural gaps close, the crypto market's vulnerability to a single-venue failure remains a defining feature of the asset class — not a residual bug from an earlier era. The AI token wave is not the cause of this vulnerability. It is the latest instrument through which the vulnerability is being expressed, in a market that has grown larger, more institutionally connected, and therefore more systemically consequential than it was when FTX fell. Readers interested in how this dynamic has played out across different speculative episodes can consult the /compare/dot-com-bubble-vs-ai-infrastructure-2024 analysis for a cross-market perspective on AI-driven concentration risk.
Why It Matters Now
History does not repeat in crypto with the mechanical precision of a textbook. But the pattern — new asset category, concentrated trading venue, inadequate custody oversight, macro stress trigger — has now produced catastrophic outcomes twice in a decade. The question is not whether exchange concentration is a risk. The evidence is unambiguous that it is. The question is whether the market's current complacency about that risk, in the middle of an AI-token-fuelled expansion phase, will prove as costly as it did in 2022. The full bubble encyclopedia offers sobering context for how that question has tended to resolve.