Hook
Last week, Bloomberg reported that Lightspeed Venture Partners is raising a $600 million secondary fund to acquire shares in OpenAI and Anthropic. This is not a blockchain story—but it is a story about liquidity, trust, and the failure of primary markets. It is a story that should terrify anyone who believes in decentralized capital formation.
I have spent the last decade watching capital markets evolve from over-the-counter desks for crypto tokens to opaque secondary pools for AI unicorns. The parallels are unsettling. In 2017, during the ICO boom, I reviewed over 40 whitepapers and identified predatory tokenomics in 30% of projects. I saw the same pattern: insiders selling their allocation to new funds at a discount, while retail waited for a listing that never came. Now, Lightspeed is doing the same for AI, but with a twist—they are buying shares of companies that have no intention of going public anytime soon. The hype burns out, but the robustness remains in the ledger.
Context
Lightspeed is a top-tier venture capital firm with a portfolio that includes Snap, Nutanix, and Grafana. In 2023, they were an early investor in Anthropic, the AI safety company behind Claude. They also participated in OpenAI’s later rounds but not as a lead. Now, they are launching a dedicated secondary vehicle called “Project Mercury” to increase their exposure to both companies. The fund will buy existing shares from employees, early investors, and other funds at a discount to the latest valuation—typically 10-30% less.
In the traditional venture world, secondary funds are a sign of maturation. They allow early backers to exit, provide liquidity to employees, and give new investors a chance to buy into proven winners. But in the context of AI, where OpenAI and Anthropic are burning billions of dollars annually on compute and talent, this secondary market signals something deeper: the primary market has failed to provide adequate liquidity. The companies are too large to IPO, too expensive for most institutional investors, and too risky for debt markets. So they rely on friends-and-family funds like Lightspeed to keep the music playing.
Core
Lightspeed’s dual bet on OpenAI and Anthropic is a hedge between two competing visions of AI. OpenAI is the consumer darling, the ChatGPT machine that captured the world’s imagination. Anthropic is the enterprise safety player, the Claude that promises to be “helpful, harmless, honest.” By buying both, Lightspeed is essentially saying: “We don’t know which horse will win, but we want to own both.” This is the same logic that drove crypto funds to buy both Bitcoin and Ethereum in 2020—a bet on the entire asset class rather than a single protocol.
But there is a deeper, more troubling implication. Lightspeed is not just buying shares; they are providing exit liquidity to early insiders. The sellers of those OpenAI shares are likely employees who joined in 2020-2021, when the company was valued at $20 billion. They are cashing out at a discount to the current $80 billion valuation, but they are still taking home a multi-million dollar payday. The fact that they are selling, rather than holding, is a signal that they do not believe the company will go public anytime soon. They are saying, “I’d rather have cash now than wait for a liquidity event that may never come.”
We audit the logic, for humans will always err. The logic here is that secondary markets are a necessary evil when primary markets are dysfunctional. In the crypto world, we solved this problem with tokenized equity and decentralized exchanges. We allowed anyone to buy and sell shares of a protocol without permission, with transparent pricing and instant settlement. The irony is that the very companies that are building the future of intelligence are relying on the most archaic form of capital formation: closed-door negotiations with a single fund.
In my experience auditing the Compound Finance governance mechanism in 2020, I saw how a well-designed token system can provide liquidity without sacrificing control. Compound’s COMP token allowed anyone to participate in governance while also providing a liquid market for speculation. The price discovery was continuous, not episodic. Compare that to OpenAI, where the only way to get exposure is to call a VC and ask for their leftover allocation. This is not a free market; it is a cartel.
Contrarian
Now, the contrarian angle. You might think that Lightspeed’s $600 million bet is a vote of confidence, a sign that AI will continue to grow. I disagree. This secondary fund is actually a sign of weakness, not strength. It shows that primary investors are desperate to exit. In crypto, we have a term for that: “exit liquidity.” The best time to buy is when everyone is selling, but here the smart money is selling to Lightspeed. The early employees of OpenAI are not stupid; they know the company’s internal metrics. If they are selling at a discount, it suggests that the growth story is losing steam.
Furthermore, the dual bet between OpenAI and Anthropic indicates that Lightspeed itself is uncertain. If they truly believed in OpenAI’s inevitability, they would not waste capital on Anthropic. They would go all-in on the winner. Instead, they are hedging, which is a classic sign of risk aversion. In the 2021 NFT market, I saw the same pattern: collectors buying both CryptoPunks and Bored Apes, hoping that one of them would survive. The result was a bloated portfolio that underperformed when the market turned.
Open source is a covenant, not just a license. Lightspeed is betting on proprietary, closed-source models. But the open-source AI movement is gaining momentum, with models like Llama 3 and Mistral closing the gap with GPT-4. If open-source models become good enough, the moat around OpenAI and Anthropic will evaporate. Then Lightspeed will be left holding shares in a commodity business.
Takeaway
We are witnessing the convergence of two worlds: the centralized capital of AI and the decentralized ethos of crypto. Lightspeed’s secondary fund is a bridge, but it is a bridge built on sand. The real question is not whether OpenAI or Anthropic will win, but whether the current capital structure can survive the next bear market. When the hype cycle ends, the only thing that will remain is the code—the logic that was audited, the protocol that was decentralized.
Hype burns out; robustness remains in the ledger. The ledger of AI is not on a blockchain, but it could be. The next wave of AI companies will be built on transparent, tokenized equity, where anyone can audit the cap table and participate in governance. Until then, we are left with secondary funds like Project Mercury, which are less about building the future and more about cashing out the present.
I seek the signal amidst the noise of the crowd. The signal here is clear: the AI capital market is broken, and the only way to fix it is to decentralize it. The question is whether the founders of OpenAI and Anthropic have the courage to embrace that future, or whether they will continue to rely on the same old tricks of the trade.
