On May 28, 2026, Anthropic announced the completion of its Series H funding round, raising $65 billion and reaching a post-money valuation of $965 billion, surpassing its rival OpenAI for the first time.

Four days later, on June 1, Anthropic confidentially submitted its S-1 filing to the U.S. SEC. Target timeline: Autumn 2026, with a potential listing as early as October. Lead underwriters have been confirmed: Goldman Sachs and Morgan Stanley.

The fundraising is expected to exceed $60 billion, potentially making it the world's second-largest IPO after SpaceX. From 'defectors' to a 'pre-IPO unicorn,' Anthropic has achieved this in just five years.

I. 2021: The 'Defection' of 11 People

That year, Dario Amodei, then VP of Research at OpenAI, left with his sister Daniela Amodei and a group of top OpenAI researchers to found Anthropic. At the time, OpenAI had not yet released ChatGPT, and the commercialization of large language models was nearly nonexistent.

No one expected this 'defector' team to surpass its former employer in just five years.

The Amodeis' departure stemmed from differing visions on AI safety. OpenAI was increasingly leaning toward commercialization, while the Amodeis believed: AI safety alignment must be embedded in the technology from its inception. They named their new company 'Anthropic'—derived from the Greek word 'anthropos,' meaning 'human.' In 2021, Anthropic had just 11 people, no products, no revenue, and only one belief: to build safe, human-beneficial AI.

II. 2023-2025: Claude's 'Slow Burn' and 'Breakout'

In 2023, Anthropic launched the Claude chatbot, which initially saw modest adoption. In 2024, Claude 3 was released, beginning to gain traction in the enterprise market. But the real turning point came in 2025.

In 2025, Claude Code debuted. This was no ordinary programming assistant—it was an 'AI colleague' capable of reading code, modifying it, running tests, checking documentation, calling tools, and generating patches. A programmer using Claude Code was like hiring a junior engineer who worked 24/7, never took leave, and required no benefits.

Data shows that as of May 2026, 80% of the programs deployed in Anthropic's codebase were autonomously generated by Claude, while R&D Engineer (This Chinese text seems out of context here and may need verification or removal) daily effective code output increased eightfold. Globally, 4 out of every 100 code commits were written by Claude.

Even more striking was the revenue growth. Just six months after Claude Code's launch, its annualized revenue surpassed $1 billion; by February 2026, it exceeded $2.5 billion. This growth rate was unprecedented in commercial software history—Salesforce took nearly five years to reach $1 billion in annualized revenue, ServiceNow took four, and Claude Code did it in nine months.

In April 2026, Anthropic released Claude Opus 4.7, offering significant improvements over its predecessor in advanced software engineering, long-duration tasks, and visual understanding while maintaining the same price. Shortly after, Claude Mythos Preview was unveiled—an unreleased cutting-edge model for cybersecurity and code vulnerability discovery, achieving 83.1% on cybersecurity vulnerability reproduction benchmarks.

Anthropic also launched Project Glasswing, collaborating with 12 top institutions, including AWS, Apple, Broadcom, Cisco, Google, Microsoft, and NVIDIA, to enhance critical software and open-source ecosystem security. The company committed $100 million in model usage credits.

III. 2026: From 'Fundraising Machine' to 'Money Printer'

In 2026, Anthropic's financials began to reshape market perceptions of the company. In Q1 2026, revenue hit $4.8 billion.

In Q2 2026, revenue is expected to reach $10.9 billion, more than doubling quarter-over-quarter, with an estimated operating profit of $559 million—its first profitable quarter.

In April 2026, annualized recurring revenue (ARR) surpassed $47 billion.

In May 2026, Anthropic completed its Series H funding round, raising $65 billion and reaching a valuation of $965 billion. From $4.8 billion to $10.9 billion in just one quarter—this wasn't growth; it was nuclear fission.

IV. Autumn IPO: Why Now?

Anthropic's choice to go public in autumn 2026 was carefully calculated.

First, the financials are impressive. Quarterly revenue of $10.9 billion and operating profit of $559 million demonstrate 'profitability,' appealing to public market investors.

Second, the market environment is favorable. In the first half of 2026, AI chip company Cerebras Systems saw its stock surge 68% on its debut day, SpaceX listed at a $1.77 trillion valuation, and CoreWeave jumped nearly 200%. The IPO window is open.

Third, the race with OpenAI. OpenAI is also preparing for an IPO, but internal disagreements between CFO Sarah Friar and CEO Sam Altman over timing may delay it until 2027. Anthropic's head start could make it the first major AI model developer to go public globally.

Fourth, funding for compute reserves. Anthropic has signed major compute partnerships with Amazon and Google, securing about 3.5 gigawatts of compute resources starting in 2027. These long-term commitments require massive funding, making an IPO the most direct financing channel.

Fifth, Dario Amodei's vision of a 'Genius Nation' requires capital. In February 2026, Amodei proposed the concept of a 'genius nation within data centers,' predicting that by 2026 or 2027, a single model's intellectual capacity would equal that of tens of thousands of top geniuses working together. Realizing this vision demands sustained compute investment and capital support.

V. The Money-Making Formula: Why Anthropic Outearns OpenAI

Anthropic and OpenAI are the 'twin titans' of AI, but their profitability gaps are widening.

The core difference lies in revenue structure. In Anthropic's ARR, 75%-85% comes from API business—billed by token usage, paying only for what's used. Subscription revenue accounts for just 15%, with consumer subscriptions at 5%. OpenAI? In Q1 2026, over 65% of its revenue came from subscriptions—ChatGPT Plus, Pro, and Team monthly/yearly fees. Consumer subscriptions made up about 40%.

API and subscription models may seem similar, but they're fundamentally different. Subscription sells 'seats'—$20 per user per month, regardless of usage. More users mean linear revenue growth; stagnant users cap revenue. API charges by volume, selling 'workflow depth'—the more agents a user deploys and tokens consumed, the higher the revenue. Revenue isn't capped by seat count but unlocked by workflow depth.

This explains Anthropic's 500% Net Dollar Retention (NDR). The same customer base spent six times more a year later.

More critically, Anthropic's inference gross margin has surged from ~38% twelve months ago to over 70%. Compute costs are relatively fixed—once models are deployed, the incremental cost of selling an additional token is nearly zero.

VI. Lessons for Workers: Anthropic's 'Comeback Formula'

Anthropic's story is legendary, but what does it mean for ordinary workers?

Lesson 1: Choosing the right 'billing model' matters more than the industry.

Anthropic chose API volume-based billing, not subscriptions. This ties revenue deeply to customer value, not linearly to customer count. For workers: Align your income with 'customer value depth,' not 'time/seat count.' Project-based revenue sharing has a higher ceiling than hourly rates; performance bonuses offer more flexibility than fixed salaries.

Lesson 2: 'Deepening existing clients' matters more than 'acquiring new ones.'

Anthropic's 500% NDR shows the same clients can deliver 6x value. This is more efficient than constantly chasing new clients. For workers: Specializing in a field, client, or skill yields compounding returns over frequent job-hopping, direction changes, or trend-chasing.

Lesson 3: Accumulate relentlessly during 'scarce windows.'

Anthropic capitalized on the scarce window for enterprise AI programming assistants (2024-2026), capturing mindshare with Claude Code. For workers: Every industry has 'scarce windows' where accumulation efficiency is 10x higher. Identify these windows and go all-in during them.

Lesson 4: Treat 'costs' as 'investments' to level up.

Anthropic doesn't view training compute as a cost but as 'capital reinvestment for future revenue and pricing power.' It even invented EBTIT (Earnings Before Training Investment) to measure how much cash inference operations generate before reinvesting in next-gen models. For workers: Don't treat learning, certifications, or networking as 'costs'—see them as 'investments.' The time, money, and effort you spend today build your future 'inference revenue.'

VII. Risks: This 'Money Printer' Isn't a Perpetual Motion Machine

Anthropic's story is thrilling, but risks loom large.

Token budget controls: Will enterprises reining in token budgets disrupt ARR growth?

Price competition: If OpenAI, Google, and Meta engage in a four-way code model battle, token prices and margins will compress.

Open-source model catch-up: If open-source models narrow the gap with closed-source frontiers, some workloads may shift to cheaper alternatives.

Compute shortages: Anthropic may not want to profit so early but is constrained by available compute, unable to reinvest all potential cash flow into expansion.

Geopolitical risks: Amodei has repeatedly cited chip supply chain disruptions or societal turmoil as major uncertainties.

Anthropic's rise is one of AI's most legendary startup stories. Five years ago, it was OpenAI's 'defector'; five years later, it's a $965 billion 'pre-IPO unicorn' preparing for an autumn IPO. But its legend isn't just about valuation and profits. It proves that in the AI era, 'profitability' matters more than 'fundraising,' 'customer depth' matters more than 'customer count,' and 'workflow embedding' matters more than 'tool provision.'

Even Dario Amodei, who left OpenAI in 2021, likely didn't foresee this outcome. But his story teaches us: Persistent accumulation in the right direction yields exponential returns over time.

When Anthropic's bell rings on Nasdaq or the NYSE in autumn 2026, it won't just be a capital feast—it'll be a 'defector's' finest answer to his original vision.

Disclaimer: This article is for financial commentary only and does not constitute investment advice. All corporate data and regulatory events mentioned are from public information and are for reference only. Official announcements shall prevail. Image source: Internet. Contact us for copyright removal if needed.

Hit 'Like,' 'Share,' and 'Heart' in one click!

Share your thoughts in the comments!

Daily AI-Powered Financial Insights