Citibank just joined the underwriting syndicate for Anthropic's IPO. The list now includes Goldman Sachs, Morgan Stanley, JPMorgan, and Citibank. This is not a routine expansion. It is a strategic escalation in the AI capital war.
In a bear market where liquidity is scarce, Anthropic is building a financial infrastructure that rivals its technical one. The move comes as the broader tech IPO window remains partially open, but only for companies with a clear narrative and a massive balance sheet. Anthropic, with $7.5 billion in total funding and a rumored valuation of $18 billion in its last private round, is now positioning for a public debut that could exceed $50 billion.
Context: The Infrastructure of an IPO
Anthropic was founded in 2021 by Dario Amodei and Daniela Amodei, both former OpenAI executives. The company's core thesis is AI safety first—building models that are aligned, interpretable, and less prone to harmful outputs. This positioning has attracted major strategic investors, including Amazon (which invested $4 billion) and Google. The company's flagship model, Claude, competes directly with OpenAI's GPT-4 and Google's Gemini.
But the path to IPO has been non-trivial. The AI industry is capital-intensive, with compute costs running into the billions. Anthropic's own estimates suggest it spends over $1 billion annually on training and inference. To sustain that burn rate, the company needs access to public markets—not just venture capital.
The addition of Citibank is a signal that the company expects a multi-billion-dollar debut that requires global distribution capacity. Goldman Sachs and Morgan Stanley are the traditional tech IPO powerhouses. Citibank brings a different network: more retail distribution, more international reach, and a strong fixed-income desk. This suggests Anthropic may issue convertible bonds alongside equity, or at least create a structure that appeals to both growth and income investors.
Core: The Financial Engineering of AI Safety
Capital congestion is the new bottleneck. In the AI arms race, the companies that win are not necessarily the ones with the best models—they are the ones with the most efficient capital structures. OpenAI, despite its mindshare, has a complex relationship with Microsoft and a non-profit governance structure that complicates its IPO prospects. xAI is still early. Anthropic, by contrast, has a clean corporate structure and a clear narrative: safety as a premium feature.
Let's break down the syndicate:
- Goldman Sachs: Lead left bookrunner. Handles the tech-heavy investor base, including hedge funds and mutual funds. Known for pricing IPOs aggressively.
- Morgan Stanley: Co-lead. Strong in retail distribution and high-net-worth individuals. Often used for consumer-facing tech IPOs.
- JPMorgan: Co-manager. Brings a massive asset management arm and relationships with pension funds.
- Citibank: Co-manager with a twist. Citibank is the most international of the four, with a strong presence in Asia and Europe. It also has a deep bench in regulatory compliance—a critical asset given the SEC's increasing scrutiny of AI disclosures. The inclusion of Citibank is a bet on regulatory arbitrage.
From my experience analyzing the FTX collapse, I know that when a company adds a major bank with a strong compliance culture, it's often a signal that the company expects heightened regulatory scrutiny. Anthropic is positioning itself as the most compliant AI company, and Citibank's risk management expertise will help navigate the SEC's new guidelines on AI-related risk factors.
Valuation Scenarios: The Safety Premium
The core valuation battle is not about model benchmarks. It's about whether AI safety can command a premium over AI capability. In the private market, Anthropic's last round valued it at $18 billion—roughly 36x its estimated 2024 revenue of $500 million. But that revenue is heavily concentrated: a few large enterprise customers (including Amazon and Google) account for the majority of API revenue. For a public market debut, the company needs to show diversification.
If the IPO prices at $30-50 billion, that would be a 60-100x revenue multiple—high, but not unprecedented for AI companies in a bull market. However, we are in a bear market for risk assets. The Nasdaq has corrected, and AI hype has cooled. The IPO will be a stress test for the AI safety thesis.
Contrarian: The Unseen Angle
Everyone is focusing on the AI technology race. But the Citibank addition reveals a different truth: Anthropic's real advantage is not its model—it's its financial engineering. The company is building a capital structure that allows it to survive a prolonged bear market and outspend competitors on compute. The IPO is a stress test for the AI safety thesis.
Meanwhile, the AI safety narrative is being monetized, not just preached. This is reminiscent of how Tether built its financial infrastructure to dominate stablecoins. The technology was secondary to the capital network. Anthropic is doing the same thing: it's using the safety narrative to attract a different class of investors—pension funds, insurance companies, and sovereign wealth funds—that are more risk-averse and care about compliance. Citibank gives them access to these investors.
Valuation is a function of narrative bandwidth. In a bear market, only the most compelling stories get funded. Anthropic's story is that it's the safe choice. Citibank's involvement is a seal of approval that this story is credible.
The Bear Market Calculus
I've been through three crypto bull cycles and two bear markets. The pattern is the same: the companies that survive are those that can access capital when others cannot. Anthropic is doing exactly that. By building a deep IPO syndicate now, it's signaling to the market that it has the financial firepower to outlast the downturn.
From my 2020 DeFi analysis, I saw how yield aggregators that raised during the crash dominated the next cycle. The same principle applies here. Anthropic's IPO prep is a strategic play to secure a war chest before the next AI winter.
But there are risks. The SEC's AI disclosure requirements are still evolving. If the company is forced to disclose details about its model's failure modes, it could spook investors. Additionally, the competitive landscape is fluid. OpenAI could launch a competing product or announce its own IPO, diluting the narrative. The IPO is a stress test for the AI safety thesis.
Takeaway
Watch for the S-1 filing. The key numbers are not the billion-dollar valuations—they are the revenue multiples, the customer concentration, and the compute cost disclosures. That will tell us if Anthropic is a real business or just a well-funded research lab. The IPO will be the ultimate stress test for the AI safety thesis. If Citibank's distribution network delivers a strong debut, expect a wave of 'compliant AI' IPOs. If not, the safety narrative loses its financial edge. The clock is ticking.