The narrative is already being written: Anthropic, the “safe AI” champion, is taking its final bow before the grand stage of the public markets. The inclusion of Citigroup alongside Goldman Sachs and Morgan Stanley for its IPO is being spun as a sign of institutional confidence. But the bubble isn't the story; the story is the story selling it. And in this case, the story is being sold to cover a structural fault line that no one wants to admit: traditional institutions don't need your public chain—or in this case, your safety-first narrative. They need a liquid, tradable asset that promises returns. Friction reveals the fault lines no one else sees.
When I first read the news that Anthropic had added a third bulge-bracket bank to its IPO roster, my immediate reaction wasn't excitement. It was skepticism. Not because of the company itself—I've respected the technical work done on constitutional AI—but because of the pattern. I've seen this playbook before. In 2020, when DeFi protocols started hiring multiple investment banks to manage their token offerings, the signal was clear: they were preparing for a liquidity event that required more than just retail speculators. They needed institutional gatekeepers to validate the narrative. The market doesn't love uncertainty. It loves narratives. And a narrative with three banks writing checks is a narrative that can be sold to pension funds.
Context: The Capital War Behind the AI Race
Let's step back. Anthropic has raised over $7 billion from investors including Amazon, Google, and Spark Capital. Its valuation is pegged at roughly $18.4 billion post-money as of its last round. That's a lot of money for a company that, by most accounts, generates less than $1 billion in annualized revenue. The math doesn't add up for a traditional IPO. Which is precisely why the bank roster matters.
Goldman Sachs and Morgan Stanley are the gold standard for tech IPOs. They have deep relationships with institutional investors. But Citigroup? That's a wild card. Citi is known for its strength in emerging markets and its ability to distribute shares to a broader, more diverse set of investors, including sovereign wealth funds and family offices. Adding Citi suggests Anthropic is preparing for a global book build—one that goes beyond the traditional US tech investor base. This is a signal that the company wants to tap into the vast pool of capital that is hungry for AI exposure but wary of the volatility.
But here's the friction: the more banks you add, the more fees you pay. And the more you dilute the premium of scarcity. A single bank can command a premium for a hot IPO. Three banks are a committee. And committees are bad at pricing risk. The signal is not just confidence; it's also a hedge against failure. If one bank can't fill the order book, the other two can. This is defensive positioning, not offensive.
Core: What the Hiring Actually Reveals
Based on my experience auditing DAO governance and tokenomics, I've learned to read between the lines of financial engineering. The announcement of an IPO bank team is itself a form of marketing. It's a signal to the market that the company is “IPO-ready.” But the substance behind that signal is often thin. Let's look at the technical details.
Anthropic's core product, Claude, is a large language model that competes with OpenAI's GPT-4 and Google's Gemini. The company's differentiation is its focus on safety and alignment. But the revenue model is still evolving. Most of Anthropic's revenue comes from API access and enterprise subscriptions. The unit economics are murky. The cost of training and inference for frontier models is astronomical. For every dollar of revenue, Anthropic likely spends two dollars on compute. This is a classic “grow at all costs” strategy that works in private markets but is punished in public markets.
Adding Citigroup is a clever move to address this. Citi has a strong track record of taking high-growth, unprofitable tech companies public—think of the 2021 IPOs of Coinbase, Robinhood, and others. They know how to tell a story of future potential. But the market of 2024 is different. Investors are demanding profitability. The “growth at all costs” narrative is dead. The question is: can Anthropic pivot to a narrative of “sustainable growth with safety premium”?
This is where the contrarian angle comes in. The dominant narrative is that Anthropic's IPO will be a slam dunk because AI is the next big thing. But the data suggests otherwise. The median IPO in 2024 has underperformed the S&P 500 by 15% in the first six months. The market is saturated with SPACs and de-SPACs that have left a bad taste in investors' mouths. The AI hype cycle is peaking. The window for a high-profile IPO is closing.
Contrarian Angle: The Unreported Blind Spot
Here's what no one is talking about: the regulatory risk that Anthropic's own safety narrative creates. The company has built its brand on being the “responsible AI” company. But with great responsibility comes great liability. When you promise safety, you open yourself up to lawsuits when something goes wrong. If Claude produces a harmful output, the legal exposure is amplified because Anthropic explicitly marketed itself as safe. This is different from OpenAI, which has always been more ambiguous about safety.
Think of it like a DeFi protocol that audits its code and markets itself as “secure.” When a vulnerability is found, the market punishes the protocol more severely than a protocol that never claimed security. The same will happen to Anthropic. The IPO will trigger a wave of scrutiny from regulators, especially in the EU and US, who will want to see exactly how Anthropic measures and mitigates risk. The cost of compliance will be enormous.
Moreover, the inclusion of Citigroup might be a red flag. Citigroup has been under regulatory scrutiny for its own risk management practices. In 2023, the bank was fined over $100 million for failing to improve its internal controls. The irony is palpable: a bank with a troubled compliance record is helping to take a “safe AI” company public. The friction reveals the fault lines: the entire ecosystem is built on narratives that don't hold up under scrutiny.
Takeaway: What to Watch Next
The next signal is the S-1 filing. When Anthropic releases its financials, we will see the true state of its business. I expect the numbers to be ugly: high revenue growth but massive losses, thin margins, and a heavy reliance on cloud credits from Amazon. The key metric to watch is the “cost of revenue” as a percentage of revenue. If it's above 80%, the unit economics are broken. The IPO will be a test of whether the market can stomach a company that loses money on every model inference.
But more importantly, watch the lock-up period and the insider selling. If the founders and early investors sell a significant portion of their shares in the IPO, it's a signal that they are cashing out. That's a bearish sign. If they hold, it's a signal of confidence. Based on the pattern of recent AI IPOs, I expect heavy insider selling.
In the end, the Anthropic IPO is not about AI. It's about capital. The companies that survive the next cycle will be the ones that can raise the most money at the highest valuation, not the ones with the best models. The bubble isn't the story; the story is the story selling it. And Anthropic is selling a story of safety that the market is desperate to believe. But the market doesn't love uncertainty. It loves narratives. And narratives have a tendency to break when the numbers come out. Friction reveals the fault lines no one else sees. Watch the S-1. The truth will be in the footnotes.
