Anthropic’s $30B Round: What the Investor List Reveals

On February 12, 2026, Anthropic closed a $30 billion Series G at a $380 billion valuation — the second-largest venture deal in history. Nine days earlier, on February 3, Claude Cowork had triggered a $300 billion software stock selloff. Capital didn’t disappear after that crash. It moved.

The same week the Goldman Sachs Software Index dropped 30% from its October 2025 highs, investors lined up to fund the company most responsible for the damage. The Anthropic funding round was 6x oversubscribed — originally targeting $10-20 billion before doubling to accommodate demand.

The headline numbers are extraordinary: $14 billion revenue run-rate, $2.5 billion from Claude Code alone, 500+ enterprise customers paying $1 million or more annually. But the deeper story is what the investor composition reveals — Microsoft hedging its OpenAI bet, Nvidia buying competitive balance, three sovereign wealth funds treating AI as national infrastructure. This isn’t a funding round. It’s a restructuring of the AI power stack.

The Anthropic Funding Round That’s Not Quite an Equity Deal

Start with the math that nobody is talking about. According to Anthropic’s official announcement, the company raised $30 billion and contractually committed $30 billion back to Microsoft for compute purchases. Amazon is separately building an $11 billion dedicated data center campus and securing a one-million Trainium chip commitment. A substantial portion of the capital raised is pre-allocated to the cloud providers who invested it.

This is infrastructure financing dressed as equity. The investors write a check, the AI lab writes it back, and both sides book it as a win — the lab gets a valuation bump, the cloud provider locks in a decade of revenue. The $30 billion headline is real. What it buys is more nuanced than the press releases suggest.

Microsoft’s $5B Hedge (and What It Means for OpenAI)

Microsoft owns roughly 27% of OpenAI’s economic interest (as of October 2025). It also just invested $5 billion in OpenAI’s most direct competitor. Crunchbase reported on the full investor list, and Microsoft’s presence is the most strategically significant entry. The explicit motive: diversify away from OpenAI dependence.

The logic is sound, if ruthless. If OpenAI dominates the AI market unchallenged, Microsoft faces two problems: antitrust exposure from its near-majority stake, and zero bargaining power when OpenAI sets pricing. Investing in Anthropic gives Microsoft a regulatory defense — “we support competition” — and a credible alternative partner if the OpenAI relationship sours. Both OpenAI and Anthropic are preparing for 2026 IPOs, and Microsoft now holds cards in both hands.

Nvidia’s $10 billion investment follows identical logic. If one AI lab dominates, it can negotiate GPU pricing down or invest in custom silicon to bypass Nvidia entirely. Two well-funded competitors keep both dependent on Nvidia hardware. Neither Microsoft nor Nvidia is buying equity upside here. They’re buying competitive balance — insurance against a future where one customer grows powerful enough to dictate terms.

Sovereign Wealth Funds Treat AI Like Oil

GIC (Singapore), MGX (Abu Dhabi), and Qatar Investment Authority all participated in this single round — three sovereign wealth funds converging on one AI company. Historically, sovereign funds reserve this commitment level for physical infrastructure: oil pipelines, deepwater ports, telecom networks. The kind of assets nation-states consider strategic.

GIC didn’t just participate — it co-led the round. GIC’s Chief Investment Officer of Private Equity, Choo Yong Cheen, called Anthropic “the clear category leader in enterprise AI.” That’s not a VC partner tweeting praise. When a sovereign wealth fund managing $770 billion in assets describes your startup that way, it signals strategic importance beyond financial returns.

These investors operate on different time horizons, different risk tolerances, and different strategic objectives than Sand Hill Road. When nation-states classify frontier AI labs alongside energy companies and telecom monopolies, it changes the competitive dynamics. AI has become infrastructure — and countries are positioning accordingly.

The Revenue That Justifies the Valuation

Anthropic’s revenue trajectory is historically unprecedented in enterprise software: $10 million (2022), $100 million (2023), $1 billion (end of 2024), $9 billion (end of 2025), $14 billion run-rate as of February 2026. That’s 10x annual growth for three consecutive years. Salesforce took 17 years to reach $14 billion in revenue. Anthropic did it in four.

The enterprise penetration tells the story behind the number. Over 500 customers now spend $1 million or more annually — up from just 12 two years ago, with customers spending $100,000+ growing 7x year-over-year. Eight of the Fortune 10 use Claude. This isn’t consumer-app growth fueled by free tiers and viral loops — it’s enterprise revenue with high switching costs and long contract cycles.

Then there’s Claude Code: $2.5 billion in annualized revenue, doubled since January 2026, now exceeding GitHub Copilot’s $2 billion ARR despite launching over three years later. It accounts for 4% of all public GitHub commits — double from the month prior — and over half its revenue comes from enterprises. ServiceNow deployed it across its entire 29,000-employee workforce.

A single product generating 18% of Anthropic’s total revenue, growing at that rate, would justify a standalone company valuation. Inside Anthropic, it’s the same growth engine behind the Cowork disruption that crashed software stocks.

At $380 billion on $14 billion in revenue, Anthropic trades at roughly 27x — high but defensible if growth sustains even partially. If the company hits its internal $70 billion revenue projection for 2028, the current valuation looks like a discount.

The February Paradox: Safety Exits, Stock Crashes, Then Record Investment

Consider the ten days leading up to the close. On February 3, Claude Cowork’s legal plugin wiped $300 billion from software market caps in a single day. On February 5, Claude Opus 4.6 launched alongside GPT-5.3-Codex and triggered another $285 billion wave of losses.

On February 9, Anthropic’s head of Safeguards Research, Mrinank Sharma, resigned publicly, warning that “the world is in peril” and that employees “constantly face pressures to set aside what matters most.” The next day, CNN reported broader safety researcher departures across OpenAI, Anthropic, and xAI.

Two days later, GIC’s Chris Emanuel praised Anthropic’s “thoughtful approach to AI development” as justification for co-leading the round. The contradiction is instructive. If safety were genuinely material to the investment thesis, losing your head of Safeguards Research three days before closing should have triggered due diligence concerns. It didn’t.

The reality is simpler: institutional investors treat Anthropic’s safety positioning as a go-to-market advantage for regulated enterprises — banks, healthcare, government — not as a substantive commitment they’re underwriting. Safety sells contracts in industries where “responsible AI” is a procurement checkbox. Whether the research behind that brand survives personnel losses is, apparently, someone else’s problem.

Still Burning Cash in 2026 (and Why Investors Don’t Care)

Despite that $14 billion revenue run-rate, Anthropic is not profitable. WSJ documents obtained by CyberNews show Anthropic expects its cash burn to drop to roughly one-third of revenue in 2026, then to 9% by 2027. Gross margins are expected to reach 50% in 2026, improving to 77% by 2028, with cash-flow breakeven targeted for 2027–2028.

For perspective, OpenAI projects $14 billion in cash burn during 2026 and breakeven not until 2030. Anthropic’s two-year head start on profitability is one reason investors favored it despite the smaller revenue base.

The early backers are already seeing returns — on paper. Amazon’s $8 billion stake has ballooned to $60.6 billion, a 7.5x return, with an additional $15 billion gain expected in Q1 2026 earnings. That markup validates the thesis for later investors: pour capital in, tolerate losses during market-share capture, and wait for the IPO exit.

Claude Cowork’s demolition of software stocks is, perversely, the strongest proof of concept for this strategy.

What This Round Really Buys

If Microsoft and Nvidia are hedging against any single lab becoming too powerful, what happens when both OpenAI and Anthropic go public in 2026 and capital markets pick a winner anyway? Infrastructure providers can invest in both privately, but public markets won’t value both equally forever. The hedge works only as long as the race stays close.

The real story in this Anthropic funding round isn’t the $30 billion or the $380 billion valuation. It’s that the companies selling infrastructure to AI labs are also the ones funding them — and they’re doing it to prevent any single customer from gaining enough power to renegotiate compute pricing. That’s not a venture capital round. That’s a cartel buying insurance.

Anthropic has hired Wilson Sonsini for IPO preparation, and Fortune reports it now ranks among the three largest IPO candidates of 2026 alongside SpaceX and OpenAI. That filing — expected in Q2 or Q3 — will force Anthropic to disclose the actual terms of its compute commitments, revealing whether this $30 billion is truly equity or structured infrastructure financing with a venture-capital label. Until then, we’re taking the headline number at face value. The S-1 will tell us if we should have been more skeptical.