I’ve been watching the OpenAI and Anthropic IPO race pretty closely — and not just because it’s a fascinating capital markets story. It’s because the outcome has real implications for how accounting and finance professionals should be managing AI vendor relationships, budgeting for AI costs, and advising clients right now.
Here’s the number that should get your attention: $1.23 trillion. That’s the combined private-market valuation of OpenAI and Anthropic as of early April 2026. Neither company is profitable. Both are planning to go public in the same quarter. And both are pricing themselves at multiples that have never existed at IPO in the history of tech.
Revenue Multiples That Have No Precedent
OpenAI’s $850 billion valuation on roughly $25 billion in annualized revenue puts them at a 34x forward revenue multiple. Anthropic at $380 billion on ~$19 billion ARR is 20x. To put that in context: Google went public at 7x in 2004, profitable. Facebook listed at 26x in 2012, also profitable — and the stock still dropped 50% in the first three months of trading. Snowflake came in at 175x in 2020, not profitable, and eventually fell about 70% from its highs.
OpenAI and Anthropic are asking public markets to accept multiples that exceed Facebook’s — without the profitability that Facebook had. Public investors read S-1s. Private investors read pitch decks. That distinction matters a lot when these filings land later this year.
The SoftBank Countdown Clock
The detail I keep coming back to is the SoftBank bridge loan. To fund its $30 billion commitment to OpenAI’s latest fundraising round, SoftBank borrowed $40 billion — unsecured — from JPMorgan, Goldman Sachs, and three major Japanese banks. Largest unsecured bridge loan in corporate history. It matures in 12 months.
Think of it like a balloon payment mortgage on a house you haven’t listed yet. Every month that passes without the IPO is a month closer to a very uncomfortable refinancing conversation. S&P shifted SoftBank’s credit outlook to negative. The banks that underwrote it wouldn’t have done so unless they had conviction the IPO happens before March 2027. That 12-month clock is running.
The Enterprise Flip Nobody Talked About
Here’s what I think is the most underappreciated part of this story. In early 2025, Anthropic represented about 10% of combined OpenAI-plus-Anthropic enterprise spending. By February 2026 — twelve months later — that number was 65%. That’s not a good sales quarter. That’s a structural shift in who enterprise customers trust.
80% of Anthropic’s revenue comes from enterprise customers. 70% of Fortune 100 companies are Claude customers. Claude Code hit $2.5 billion in annualized run-rate revenue just nine months after launch. Meanwhile, OpenAI still leads on consumer metrics — 900 million weekly active users is a real number — but consumer engagement and enterprise revenue are different animals. Enterprise contracts generate predictable cash flows. That’s what DCF models are built on, and that’s what public market underwriters weight most heavily.
What This Actually Means for You
Two things I want you to take away from this.
First: your current AI pricing is below cost. Both OpenAI and Anthropic are subsidizing today’s pricing with investor capital. OpenAI doesn’t expect to be profitable until 2029 or 2030. Anthropic is targeting 2028. Until then, the subsidy continues. But the moment these companies answer to public shareholders, the calculus changes. Quarterly earnings pressure doesn’t tolerate indefinite subsidies. The customers who absorb the resulting margin expansion will be enterprise clients — which means you and your clients. I’d model for 30–50% inference cost increases over the next 24 months and build that into your AI budget projections now. If you’re advising clients locking into multi-year AI contracts, this is a conversation worth having before they sign.
Second: vendor concentration is now a risk management question. Think about what happened with Salesforce post-IPO. Companies that had locked into multi-year enterprise contracts before the listing found themselves renegotiating at significantly higher rates as Salesforce optimized for ARR growth and margin over customer flexibility. The incentive structure changed the moment they went public. With 70% of the Fortune 100 running on Anthropic’s platform, that kind of concentration is no longer just a technology architecture decision — it’s a vendor risk exposure. Multi-vendor AI strategy is becoming what multi-bank relationships and multi-cloud infrastructure already are: a risk management default, not a preference.
Key Takeaways
- OpenAI (34x) and Anthropic (20x) are targeting forward revenue multiples that exceed every comparable tech IPO — without profitability. The closest comps didn’t go well for early investors.
- SoftBank’s $40B unsecured bridge loan maturing March 2027 is a structural countdown clock on the OpenAI IPO. If it slips, that refinancing conversation gets expensive fast.
- Your current AI pricing is subsidized. Build in 30–50% cost increases over 24 months. Don’t let clients lock into long-term contracts at today’s rates without modeling the post-IPO scenario.
- Post-IPO, vendor incentives shift. Pricing flexibility, contract terms, and roadmap alignment all change when a company answers to quarterly earnings. Multi-vendor AI architecture is now a risk management imperative.
- Anthropic’s 10% → 65% enterprise share shift in 12 months is a structural change in the market. If you’re helping clients think through AI vendor strategy, this trajectory matters more than the headline valuations.
Want the CPE credit? Take the full lesson on EverydayCPE and earn 0.2 CPE credits: [lesson link]
Leave a Reply