The total value locked in decentralized finance protocols had just crossed $50 billion in late 2023, a tentative recovery after the Terra collapse. By August 2024, that number had stagnated, oscillating around $48 billion, while the combined implied valuation of OpenAI and Anthropic had doubled. In the same week, Lightspeed Venture Partners, a firm that had conspicuously avoided direct crypto asset exposure, announced a $600 million secondary fund to acquire shares of both companies. The juxtaposition is not coincidental: it is a cartographic representation of where institutional liquidity is flowing, and where it is not.
Context: The Secondary Fund as a Macro Instrument
Lightspeed’s plan, first reported by Bloomberg and later filtered through Web3 news aggregators, involves raising a dedicated secondary vehicle—Project Mercury—to purchase existing shares in OpenAI and Anthropic from early employees, angel investors, and perhaps other institutions. Unlike a traditional venture fund that deploys capital into primary rounds, a secondary fund buys already-issued equity, providing liquidity to sellers who have been locked in for years. This is a departure from the standard VC playbook, and it signals a shift in capital allocation strategy: from “discovering unicorns” to “holding unicorns.”
The macro context is critical. The US Federal Reserve had maintained elevated interest rates throughout 2024, compressing risk appetite across asset classes. Crypto, despite the January approval of spot Bitcoin ETFs, failed to attract the wave of institutional capital that many had predicted. The ETF inflows, while significant in absolute terms, were largely recycled from existing crypto-native capital rather than representing new money from pension funds or endowments. Meanwhile, the AI sector saw a flood of investment, with OpenAI and Anthropic raising billions in primary rounds at valuations exceeding $80 billion and $20 billion, respectively. Lightspeed’s secondary fund sits at the intersection of these two trends: it is a bet that AI will continue to outperform, and a tacit admission that crypto’s liquidity story is stalling.
Core: The Structural Integrity of Crypto’s Liquidity Model
Based on my own modeling of institutional flows over the past three years—a framework I built after the 2022 Terra collapse, when I spent two months mapping the interconnections between stablecoin minting, exchange reserves, and DeFi yields—I have come to view secondary funds as a canary in the liquidity mine. When a top-tier venture firm like Lightspeed shifts its focus from primary to secondary markets, it is not just a tactical move; it is a structural acknowledgment that the primary market has become too frothy, too illiquid, or both. In the AI sector, the primary market is indeed frothy, with rounds oversubscribed and valuations decoupled from revenue. But the secondary market offers a discount: sellers are often willing to accept a 10–30% discount to the latest valuation to convert paper into cash. Lightspeed is essentially arbitraging this gap.
The same dynamic is playing out in crypto, but with a twist. Crypto’s secondary market—the exchanges where tokens trade—is far more liquid than any private secondary market for AI equity. However, the liquidity is shallow and concentrated. During the 2024 sideways market, I observed that the top 10 tokens accounted for over 80% of global exchange volume, while the rest of the market experienced a liquidity drought. The chaotic surface of crypto price action masks a deeper structural vulnerability: the reliance on a small set of highly speculative assets to provide the illusion of a liquid market. When institutional capital compares the two, the choice becomes clear. AI offers a game with a finite number of players, clear moats, and a path to profitability, however long. Crypto offers an infinite game of protocol competition, regulatory uncertainty, and fragmented liquidity.
The Ethical Vulnerability of the Dual Bet
Lightspeed’s simultaneous investment in both OpenAI and Anthropic is a hedge, but it is also a confession. The cold algorithmic data of secondary market pricing—the discount rate, the lock-up period, the implied volatility—sits uncomfortably against the ethical promise of decentralized intelligence. OpenAI and Anthropic are both centralized, permissioned systems. Their governance is opaque, their decision-making concentrated in small teams, and their business models dependent on the extraction of value from user data. By doubling down on both, Lightspeed is implicitly arguing that the future of AI will be a duopoly, not a decentralized network. This is philosophically disillusioning for anyone who believed that the blockchain ethos of trustless, open systems would extend to the next wave of technology.
And yet, there is a deeper irony. The very liquidity that Lightspeed is providing to OpenAI and Anthropic’s early investors is a form of decentralization: it allows capital to flow from locked, concentrated positions to a broader set of holders. But it is a decentralization of ownership, not of control. The power to decide the direction of the technology remains with the founders and their boards. In crypto, the opposite is often true: control is supposed to be decentralized through governance tokens, but ownership is concentrated in the hands of early investors and team wallets. The ethical vulnerability lies in the mismatch between the narrative and the reality. Lightspeed is not acting in bad faith; it is simply responding to market incentives. But the result is a reinforcement of the very structures that crypto was supposed to disrupt.
Macro-Historical Synthesis: The Rhythm of Capital
This is not a new story. It is the same rhythm of capital that has played out in every technological revolution since the railroads. In the late 1990s, as the internet bubble inflated, venture capital shifted from early-stage bets to late-stage, mezzanine-type funds that provided liquidity to insiders. The pattern was the same: a hot sector attracts capital, valuations rise, early investors seek exits, and secondary markets emerge to facilitate those exits. When the bubble burst, the secondary funds were often the first to suffer, because they had bought at inflated prices. But the survivors—the firms that had bought at a discount—emerged stronger.
If we apply this historical lens to the current moment, the Lightspeed fund is a signal that the AI sector is entering the late-stage of its cycle. The next phase, if the pattern holds, will be a correction, followed by consolidation. The question for crypto is whether it will be swept up in the same correction, or whether it will decouple. The decoupling thesis—that crypto will rise independently of traditional markets and AI in particular—has been a popular narrative among crypto maximalists. But the evidence from the past 18 months suggests otherwise. When the Fed hiked rates in 2023, both crypto and AI stocks fell. When the AI hype accelerated in 2024, crypto remained flat. The correlation is not perfect, but it is positive. The two asset classes are competing for the same pool of speculative capital, and right now, AI is winning.
Contrarian: The Decoupling Thesis Is Fragile, But Not Dead
The contrarian view is that this capital rotation is cyclical, not structural. AI is a technology that requires massive centralized infrastructure; crypto is a technology that enables decentralized value exchange. They are not substitutes but complements. In the long run, the argument goes, AI will need crypto for trustless data provenance, decentralized compute markets, and micropayments. Lightspeed’s bet on centralized AI is a bet on the current paradigm, but the next paradigm could be built on blockchain rails. If that happens, the capital that flows into AI today will eventually find its way into crypto infrastructure.
But this is a long-term thesis, and it ignores the short-term liquidity dynamics. The $600 million that Lightspeed is deploying into OpenAI and Anthropic is $600 million that is not going into crypto funds. Furthermore, the success of the fund may encourage other firms to follow suit, creating a virtuous cycle for AI and a vicious one for crypto. The structural integrity of crypto’s liquidity model is fractured by its own fragmentation: dozens of Layer 2s, each with its own token and user base, but the same small pool of active users. Lightspeed’s dual bet is a reminder that capital prefers concentration over fragmentation. The most valuable companies in the world are those that can aggregate liquidity and attention. Crypto, in its current state, is doing the opposite.
Takeaway: The $600 Million Question
The $600 million question is not whether Lightspeed’s bet pays off—the odds are in its favor, given the discount it is likely receiving. The question is whether the capital that flows into AI will eventually cycle back into crypto as the infrastructure for a new, decentralized digital economy, or whether the two asset classes will remain in a zero-sum struggle for the attention of the same liquidity. The next twelve months will tell us whether we are building a new financial system, or just recycling the old one. Based on my analysis of the Bitcoin ETF flows and the subsequent lack of institutional follow-through, I am leaning toward the latter. But that is the nature of a sideways market: it forces you to sit with the uncertainty, and to recognize that the most important data points are often the ones that are not yet visible.
