For the first time in the history of this market, the most valuable private AI company in the world isn’t OpenAI. According to a Financial Times report carried by Investing.com, Anthropic has agreed terms on a $30 billion Series H at a $900 billion pre-money valuation, vaulting past OpenAI’s most recent $852 billion private mark with terms expected to close before the end of May.

That number is doing a lot of work. Three months ago, Anthropic’s Series G closed at a $380 billion post-money valuation. The new round nearly triples that figure. In the same window, annualized revenue went from $9 billion at the end of 2025 to north of $45 billion — a 5x curve in under five months, almost entirely on the back of Claude Code enterprise adoption.

The capital markets are placing a directional bet. The bet is on infrastructure, not chat.

Quick Summary: Anthropic’s Series H at a Glance

Detail	Info
Round size	$30 billion
Pre-money valuation	$900 billion
Co-leads	Sequoia Capital, Dragoneer Investment Group, Greenoaks Capital, Altimeter Capital
Minimum check from each co-lead	$2 billion+
Expected close	End of May 2026
Prior round (Series G, Feb 2026)	$30B raise at $380B valuation (post-money)
Reported ARR	$45 billion+ (up from $9B at year-end 2025)
Inference gross margin	70%+ (up from ~38% a year ago)
OpenAI’s most recent private valuation	$852 billion
IPO window targeted	October 2026
Primary source	FT report via Investing.com

Bottom line: A $900B private valuation isn’t just a sticker. It’s institutional capital betting that Anthropic’s developer-and-enterprise stack scales harder than OpenAI’s consumer-plus-platform stack from here. The IPO window in October will tell us whether the public market agrees.

---

What Actually Happened

Per the FT report carried by Investing.com and the Yahoo Finance writeup, Anthropic and its co-leads have agreed terms on a $30 billion Series H. Sequoia, Dragoneer, Greenoaks, and Altimeter are each on the hook for at least $2 billion, with additional investors expected to round out the allocation before the deal closes at the end of May.

The pre-money valuation is $900 billion. Post-money sits north of that. For comparison:

• OpenAI’s most recent private valuation, set by the $122 billion fundraise closed April 1, 2026, was $852 billion post-money.
• OpenAI’s $6.6 billion employee tender from late 2025 priced the company at $500 billion.
• Anthropic’s Series G (Feb 2026) closed at a $380 billion post-money valuation.

So the gap to OpenAI didn’t just close. It flipped. And it flipped on the back of a single fundraise that nearly tripled the company’s price in 90 days. The Decoder’s coverage frames it as Anthropic “approaching $1 trillion.” On the post-money math, that’s not hyperbole. It’s an arithmetic observation.

Why the Capital Markets Just Reranked the AI Race

The valuation jump only makes sense if you look at what the underlying numbers did in the same window.

ARR. Anthropic ended 2025 at roughly $9 billion in annualized revenue. By February 2026 it was at $14 billion. By April it had cleared $30 billion. The FT report now puts ARR above $45 billion. That’s a 5x curve in under five months, against a base most companies would consider a steady-state plateau.

Gross margin on inference. Per Semianalysis’s AI Value Capture report, Anthropic’s inference gross margin moved from about 38% a year ago to over 70% today. That’s the difference between “promising AI lab with terrible unit economics” and “infrastructure business with software-like margins.” The valuation re-rate is mostly that line.

Concentration risk. Nearly all the growth is attributable to Claude Code and the enterprise developer surface that wraps it. Claude Code went generally available in May 2025, hit $1B annualized by November 2025, and crossed $2.5B by February 2026. The same product is now anchoring an enterprise stack used at Netflix, Spotify, KPMG, L’Oréal, and Salesforce. The Ramp AI Index crossover earlier this month put numbers behind the anecdotes.

That’s the bull case in three lines. It’s also the risk case, which we’ll come back to. A single product driving the curve looks great in a pitch deck and dangerous in a 10-K.

The Three Bets Inside This Valuation

A $900B private mark isn’t a vote of confidence in Anthropic the company. It’s a vote of confidence in three specific bets the company is making simultaneously.

1. The bet that coding agents win the enterprise

The single most-discussed line in any Anthropic deck right now is the per-engineer monthly API spend at large customers — $500 to $2,000, according to VentureBeat’s reporting on Uber’s CTO disclosure. That’s the number that translates “engineers like the tool” into a quarterly P&L line, and it’s the number that has institutional investors writing $2B+ checks.

The bull case: every Fortune 500 engineering org is on the same Uber curve, with adoption moving from 32% to 84% in four months. If even a fraction of that adoption sustains, the revenue base under Anthropic is in the high tens of billions on coding agents alone, and that’s before any vertical expansion. Our coding-assistant comparison walks through why Claude Code consolidated the workflow that GitHub Copilot pioneered.

2. The bet that capacity gets paid for

A 70% gross margin on inference is impressive. Sustaining it requires the $40 billion Google compute deal and the matching Amazon commitments to deliver on schedule. Dedicated capacity comes online over multi-year horizons, not multi-week ones, and every “Claude is rate-limited” moment is a competitive opening.

Sequoia, Dragoneer, Greenoaks, and Altimeter aren’t betting that the capacity is already there. They’re betting that $30 billion of fresh capital, on top of existing commitments, buys enough runway to ride the demand curve until the chips and the data centers actually exist. That’s a longer-dated bet than the revenue line suggests.

3. The bet that the IPO window holds

The October 2026 IPO target has been an open secret in the market for months. The Tech Portal reported in March that bankers were modeling a raise north of $60 billion against the Series G valuation. With the Series H now at $900B, the IPO math gets more interesting — and more pressured.

Pricing an IPO above a $900B private mark is not impossible, but it requires a market that wants to absorb tens of billions of new AI exposure at frontier valuations. October is a long way away in capital-markets weather. The Series H buys Anthropic the option to delay if the window closes, but it also raises the bar for what a successful listing looks like. A flat-to-private offering at this valuation would be read as a soft landing. A premium would be read as validation. Either outcome reshapes the AI lab competitive frame for 2027.

How Anthropic and OpenAI Stack Up at Today’s Marks

The valuation crossover is the headline. The composition of how each company gets there is the story.

	Anthropic (May 2026)	OpenAI (most recent)
Latest private valuation	$900B pre-money (Series H, FT)	$852B post-money (Apr 1, 2026 raise)
Most recent round size	$30B Series H	$122B (reported)
Reported ARR	$45B+	$20B+ (per industry trackers)
Revenue mix	Enterprise / developer (Claude Code dominant)	Consumer ChatGPT + API + enterprise
Inference gross margin	70%+ (Semianalysis)	Not publicly broken out
Distribution strength	Inside developer tools, AWS, Google Cloud, Cowork	Microsoft channel, ChatGPT direct, GPT Apps
IPO target	October 2026	No confirmed timeline

Two stacks, same destination, different shapes. OpenAI’s lead historically came from consumer mindshare and the Microsoft channel. Anthropic’s lead is being built from the developer surface and the per-token economics underneath it. The capital markets just decided that, at today’s growth rates, the enterprise-and-developer stack scales faster from here. That decision is reversible. It’s also the most expensive reversal anyone in this market has ever had to fund.

The Risk Case That Doesn’t Get Enough Airtime

The bull case is loud. The risk case deserves equal time, because at $900B it determines whether this round looks brilliant or overpriced in 18 months.

Single-product concentration. Strip Claude Code out of Anthropic’s revenue base and the curve looks much closer to OpenAI’s flat 2025. The investors writing $2B checks know this. They are betting that the Claude Code curve doesn’t break — and that “Claude Code” is a category Anthropic owns, not a product OpenAI can clone with a GPT-5.5 plus Codex push. The Anthropic-vs-OpenAI competitive frame has been mostly developer-tool tilt for two quarters. A successful coding-agent counter-move from OpenAI compresses the multiple fast.

Token-pricing exposure. The customer budget conversations starting at Uber-class accounts are real. Per-engineer spend running $500-$2,000/month forces FP&A involvement, which forces contractual price caps, which forces the runaway revenue curve to flatten. The next round of enterprise renewals will price this in. The valuation assumes most of those negotiations go well for Anthropic. Many won’t.

Capacity execution risk. The bull case requires the Google and Amazon capacity deals to deliver. Any slip turns into rate limits, which turn into customer churn, which turn into a multiple compression event right as the company tries to price an IPO. Anthropic’s own capacity moat episode earlier this year telegraphed how thin the operational margin is.

Public-market reception. A $900B private mark sets a high floor for the IPO. If macro turns or AI enthusiasm cools by October, Anthropic prices flat or below — and a flat-to-private debut at this valuation is a hard story to tell shareholders. The Series H buys runway to wait, but the cap table now contains investors who will want a liquid mark at this number, not a haircut.

None of these break the company. All of them cap the upside. At $900B, the multiple is priced for execution that has to be very close to flawless.

What This Means for Buyers, Builders, and Operators

The financing news is interesting on its own. It also reshapes practical decisions for anyone making AI vendor calls right now.

1. For enterprise buyers

The Series H is a stability signal. Anthropic just locked in enough capital to fund another two years of the current burn rate, plus the IPO runway. Procurement teams who were modeling Anthropic as a platform risk against a Microsoft/OpenAI alternative can downgrade the bankruptcy line in their vendor-risk worksheet. The remaining platform risk is product direction and pricing — neither of which the funding fixes.

The other read: pricing discipline gets harder. A vendor with $30B fresh on the balance sheet has less reason to cut deals on the way into the next renewal cycle. Lock in multi-year pricing now if you have the volume to negotiate, or expect the next renewal to be at full list.

2. For developers and platform builders

Building on Claude APIs is now meaningfully less risky than it was 90 days ago. The capacity story still has friction — rate limits and tier ceilings haven’t gone away — but the existential question of whether Anthropic is around to maintain its API surface in 2027 is settled. That changes the build/buy math for anyone weighing a custom-fine-tune path versus an Anthropic API path.

It does not change the model lineup question. Pick the model that fits your workflow now, plan to revisit each major release. The funding doesn’t make Claude better at your specific task.

3. For people watching the IPO

The Series H sets the floor. The October target sets the date. Between now and then, three things drive whether the IPO lands well: the ARR trajectory (does $45B become $60B by Q3?), the gross margin trend (does 70% hold under capacity pressure?), and the competitive response (does OpenAI ship a credible coding-agent counter-move?). All three are knowable by September. None of them are decided yet.

The broader AI lab IPO speculation around Anthropic, OpenAI, and SpaceX is real, but Anthropic is the one with a concrete window. That matters for benchmark-setting. Whatever the public market pays for Anthropic in October becomes the comp for every AI company filing in 2027.

Frequently Asked Questions

Q: Why is Anthropic’s valuation suddenly $900 billion when the Series G closed at $380 billion three months ago? A: Two reasons. Revenue went from $14B annualized in February to $45B+ in May — a 3x in a single quarter, almost entirely from Claude Code enterprise adoption. And inference gross margin moved from ~38% to 70%+, turning Anthropic into a software-margin business instead of a high-burn infrastructure bet. The new mark prices those two changes in.

Q: Is Anthropic really worth more than OpenAI now? A: At private-market marks, yes — $900B pre-money versus OpenAI’s most recent $852B post-money. But “worth more” in private markets is a function of the last check written, not a steady-state valuation. OpenAI hasn’t raised since March; if it priced a round today, the mark could move. The capital markets have ranked Anthropic ahead on the most recent data, not declared a permanent winner.

Q: Who’s leading the round? A: Sequoia Capital, Dragoneer Investment Group, Greenoaks Capital, and Altimeter Capital are the four co-leads, each committing at least $2 billion, per the FT report. Additional investors are filling out the balance of the $30B raise.

Q: Does this mean Anthropic is profitable? A: No. Gross margin and net margin are different conversations. Inference gross margin at 70%+ means the unit economics on serving a model query are healthy. The company is still spending heavily on training, capacity commitments, hiring, and the PwC and field GTM motions that are driving the adoption curve. Profitability is a 2027 conversation at the earliest.

Q: When is Anthropic’s IPO? A: The target window is October 2026 per Tech Portal reporting. No S-1 has been filed yet, no ticker chosen, no exchange confirmed. Goldman Sachs, JPMorgan, and Morgan Stanley are reportedly in underwriter discussions. Treat October as a planning date, not a calendar lock.

Q: What’s the biggest risk to this valuation holding? A: A successful OpenAI counter-move on coding agents before the IPO window. Claude Code is doing nearly all of the work on the revenue curve. If GPT-5.5 plus a refreshed Codex plus aggressive pricing closes the developer-tool gap by Q3, the growth multiple compresses and the IPO prices lower. That’s the scenario the Series H investors are betting against.

Q: How does this compare to OpenAI’s last raise? A: OpenAI’s $122B raise closed April 1, 2026 and set a $852B post-money valuation. Anthropic’s Series H is half the size ($30B vs. $122B) but at a higher mark ($900B vs. $852B), reflecting different growth profiles. OpenAI’s revenue base is larger in aggregate; Anthropic’s growth rate from a smaller base is faster. The capital markets are pricing growth rate ahead of base today.

Our Take

The Series H is the cleanest signal yet that the AI lab race has split into two competing theses, not one. Through 2024 and most of 2025, the framing was “OpenAI is winning, the rest are catching up.” That framing is dead. There are now two stacks at frontier valuations, and the capital markets believe Anthropic’s developer-and-enterprise stack scales harder from here than OpenAI’s consumer-and-platform stack. That’s a real change in how this market is priced.

The part I’d push back on is the “$900B is the new floor” reading. Private marks at this scale are a function of the last check written. The Series H is locked in. The IPO mark in October is what matters for whether $900B becomes durable. If Anthropic prices the public offering at a premium, this valuation gets validated and the Mythos-class capability tier becomes the bridge to a trillion-dollar private mark. If it prices flat or below, the Series H investors take a paper haircut and the AI lab thesis gets a public-market reset. Both outcomes are live.

For OpenAI, the strategic clock just got louder. Every month between now and October is a month to ship a coding-agent counter-move that closes the Claude Code gap. The OpenAI superapp strategy is the consumer-side answer. The enterprise-side answer hasn’t shipped as a coordinated push. The window to land it before Anthropic’s IPO is finite.

For the rest of the market — investors, buyers, developers, builders — the practical move is to stop modeling this as a one-horse race and start modeling it as two stacks with overlapping but distinct economics. The crossover is real. The reversal is possible. The next six months decide which read holds.

---

Last updated: May 19, 2026. Sources: FT report via Investing.com · Yahoo Finance · The Decoder · VentureBeat · The Tech Portal · TechCrunch (OpenAI raise) · Crunchbase (OpenAI tender).

Related reading: Anthropic Just Beat OpenAI in Business Spending · Anthropic vs OpenAI in 2026 · Google’s $40B Anthropic Bet · OpenAI’s $122B Superapp Strategy · Cursor vs Claude Code vs Copilot · Anthropic Claude Mythos Leak