Five Years Ago They Walked Out of OpenAI: Thus Began The $1.4 trillion Anthropic Journey

Athropic began as a safety-focused offshoot of OpenAI. A series of products, led by a coding tool developers adopted quickly, has turned it into the largest IPO in history. The growth is extraordinary, and so are the losses and the spending commitments.

In May 2021, a few months after its founders left OpenAI, Anthropic was valued at $550 million. This week it is asking public investors to value it at up to $1.404 trillion, roughly 2,550 times that first figure (our calculation).

The company plans to sell 825 million shares at $70 to $78 each on Nasdaq, under the ticker ANT, raising up to $64.35 billion, or about ₹6 lakh crore (our calculation). That is more than 20 times the size of Hyundai Motor India’s ₹27,870 crore issue, India’s largest IPO.

Seven strategic partners, including Google, Nvidia, Microsoft and TSMC, will buy another $26 billion of non-voting shares at the IPO price. Pricing is expected on Wednesday evening US time, and trading should start on Thursday.

The founder

Dario Amodei is a scientist, which is unusual among technology founders.

Born in San Francisco in 1983, he studied physics at Stanford and earned a PhD in biophysics from Princeton, researching how neural circuits work. After a stint at Baidu, he joined OpenAI in 2016 and rose to vice president of research.

In 2021, he left with his sister Daniela and several senior colleagues over what they described as differences in direction.

Their view was that AI would become very powerful very quickly, and that whoever built it should put safety research at the centre. Dario became chief executive and Daniela president.

That view still shapes the company. Anthropic is a public benefit corporation, and its Responsible Scaling Policy sets risk thresholds that must be met with new safeguards before more powerful models are released.

Amodei has written about both sides of AI: its potential to transform medicine and science, and risks ranging from bioweapons to large-scale job losses.

He has also turned down opportunities. In 2023, during OpenAI’s boardroom crisis, he was approached about replacing Sam Altman and about merging the two companies, and declined both. In February 2026, he refused to drop Anthropic’s ban on Claude being used for mass domestic surveillance or fully autonomous weapons.

The Pentagon labelled the company a supply-chain risk, and a federal judge later temporarily blocked that action. Forbes puts his net worth at about $15.5 billion, and the IPO will raise it considerably.

Anthropic Products 

When Claude launched in March 2023, ChatGPT already had a four-month lead with consumers. Anthropic chose not to fight for consumers. It focused on businesses and developers, the customers who pay the most for AI.

The first step was models good enough for serious work. The Claude 3 family, launched in March 2024, came in three sizes, Haiku, Sonnet and Opus, so customers could choose between speed, cost and capability. Claude 3.5 Sonnet, released that June, became popular with programmers.

Next came AI that could act rather than just answer. In late 2024, Anthropic showed Claude operating a computer by reading the screen, clicking and typing. It also released the Model Context Protocol, an open standard for connecting AI to other software. Rivals adopted it widely.

Then came Claude Code, the product that changed the company’s finances. It was released in 2025 and runs in a developer’s terminal. It does more than suggest code: it reads an entire codebase, plans changes, edits files, runs tests and fixes its own mistakes.

Developers adopted it quickly, and companies bought it for entire engineering teams. Industry reports put its annualised revenue in the billions of dollars within about a year.

In January 2026 came Cowork, which offered office workers the same kind of capability Claude Code gave programmers. Inside the Claude desktop app, it could work through a user’s files and apps to finish multi-step jobs, such as organising documents, building a spreadsheet from receipts or drafting a report from notes.

It needed no coding, so lawyers, analysts and finance teams took to it quickly. When Anthropic added plugins for legal, finance and sales work, software and IT-services stocks sold off sharply, as investors worried that AI agents could replace paid software subscriptions and outsourced work. Cowork’s capabilities have since been folded into the main Claude app.

New models have arrived faster since then. A restricted, more powerful tier called Mythos was previewed for cybersecurity work in April. Mythos 5 and Fable 5 were released in June, then briefly suspended to comply with US export controls before access was restored on 1 July.

Around the models, Anthropic has built a wider set of products: Claude Science for research labs, Claude Security for finding software vulnerabilities, Claude in Chrome, Claude for Microsoft 365 and Google Workspace, and Claude Tag inside Slack.

Anthropic began by selling a model. It now wants to be where office work gets done, and that ambition is what the trillion-dollar valuation is based on.

The race to a trillion

OpenAI is six years older and better known. Anthropic’s valuation rose fastest after Claude Code took off: $61.5 billion in March 2025, $183 billion in September 2025, $380 billion in February 2026 and $965 billion in May 2026.

At the top of the IPO range, it would be worth more than OpenAI, which was reportedly valued at $1.25 trillion in its latest round.

Revenue grew at a similar pace. Third-quarter revenue was $17 billion, more than 11 times the figure a year earlier. The annualised run rate reached about $81 billion by the end of October, up from about $71 billion a month earlier. Customers paying more than $100,000 a year increased from about 950 to more than 11,900 in twelve months. No single customer accounts for more than 10% of revenue.

But its costs are also very large.

Losses. Anthropic lost $104 billion in the first nine months. Most of that, $94.3 billion, was a non-cash revaluation charge required by accounting rules. Even excluding that charge, the company is not yet profitable. An operating profit of $974 million in the second quarter turned into an operating loss of $567 million in the third.

Commitments. The company has signed up for about $900.9 billion of computing and infrastructure spending, against nearly $132 billion in cash and short-term investments. Every new model and agent needs more chips and data centres.

Valuation. At $1.404 trillion, investors would be paying about 17 times the current revenue run rate (our calculation). On banks’ 2028 revenue forecasts of $270–280 billion, reported by Reuters, the multiple falls to about five times. One investor who studied the deal and decided not to invest has a 2028 estimate 30% below the banks’. The valuation depends heavily on that forecast.

Control. Founders hold 50.1% of the voting power through a special share class, and a Long-Term Benefit Trust elects a majority of the board. Public shareholders will have very little say over how the company is run.

The prospectus also warns that using AI to build more capable AI could pose serious safety risks if progress speeds up sharply. That kind of warning rarely appears in an IPO document.

Demand for the public offer looks strong. Reuters reported that the order book was six times oversubscribed within days of the roadshow starting. Investors’ main questions are whether Anthropic can stay ahead as Google catches up, whether its office-work products can match Claude Code’s revenue, and whether it can keep charging premium prices.

The timing is difficult. Last week’s AI sell-off, the turmoil at Nvidia and Monday’s 6% fall in Taiwan’s market have unsettled investors. A strong debut would help the whole AI trade, and a weak one would increase doubts about AI valuations.

Where it would rank

At the top of its range, Anthropic would sit just behind Tesla among the world’s most valuable companies (read: The trillion dollar club here). It would be by far the youngest company at that level. SpaceX took more than two decades to reach a similar valuation. Anthropic has taken about five and a half years.

India is Claude.ai’s second-largest market after the US, with about 7% of global usage, and Anthropic opened a Bengaluru office in February.

For Indian IT companies, the listing raises an uncomfortable question. The sell-off that followed Cowork showed that investors see AI agents as a direct threat to the outsourcing model.

A company built on those agents is about to be valued at roughly four times TCS, Infosys, HCLTech and Wipro combined. Indian IT firms will need to show they can earn money by helping clients use AI, rather than losing work to it.

For foreign money flows, the effect could go either way. A strong debut could pull money back towards AI-heavy markets such as Taiwan and Korea. A weak one could help India, which has little AI exposure.

Claude Code showed that Anthropic could build something businesses would pay heavily for. Cowork suggested it could do the same beyond software. Now it may be

About The Author

Recommended For You

About the Author: Team Finance Intellect