The AI Money Race 2026: Revenue, Funding and Valuations of Anthropic, OpenAI, Mistral and Manus

Seven months ago Anthropic's revenue run rate was about $9 billion. By the end of July it was $65 billion. In the same period its valuation rose from $183 billion to $965 billion, OpenAI raised the largest private round in history ($122 billion), Mistral closed Europe's biggest tech equity round and Manus came back from a cancelled takeover with half a billion dollars. Anthropic is now preparing a Nasdaq listing that could value it at around $2 trillion. This article puts the numbers side by side and explains what the money race means for companies in Croatia and the DACH region that build on Claude, ChatGPT, Mistral or Manus.

The short answer: two giants, one European challenger, one agent specialist

The frontier AI market has split into tiers. Anthropic and OpenAI each raise tens of billions per round, are valued near or above $1 trillion and report run-rate revenue in the tens of billions. Mistral is the only European lab with comparable ambitions, at roughly a fortieth of their valuation. Manus shows what a focused agent company is worth: a few billion dollars. For buyers, the size of these numbers is less important than what they say about pricing power, vendor stability and how long today's subsidised prices will last.

Anthropic: the fastest revenue ramp in software history

Anthropic's growth is the story of the year. According to figures the company shared with investors and the press, its annualised run rate went from roughly $9 billion at the end of 2025 to $14 billion in February, $19 billion in March, $30 billion in April, $47 billion in May and $65 billion at the end of July, roughly seven times higher than a year earlier. Most of it comes from businesses: API usage, Claude Code and enterprise contracts rather than consumer subscriptions.

Two caveats matter. First, a run rate is not revenue: it extrapolates the latest month to a full year. Reuters, citing the leaked draft prospectus, reported that Anthropic actually booked $4.6 billion of revenue in 2025 with a net loss of $42 billion. Second, Anthropic counts sales of Claude through cloud partners (Amazon Bedrock, Google Vertex AI) in its revenue, which OpenAI does not. That accounting difference explains much of the debate about which lab is "bigger".

From $61.5 billion to $965 billion in 15 months

The valuation followed the revenue. Anthropic was valued at $61.5 billion in March 2025, $183 billion in September 2025 and $380 billion in February 2026. On 28 May 2026 it closed a $65 billion Series H at $965 billion post-money, led by Altimeter Capital, Dragoneer, Greenoaks and Sequoia Capital, including $15 billion of previously committed investment. That made it the most valuable private AI company, ahead of OpenAI.

The IPO: Nasdaq, probably before Thanksgiving

Anthropic confidentially submitted a draft registration statement to the SEC on 1 June 2026. Reports in early October said the listing slipped from October to November, on Nasdaq, with investors eyeing a valuation of around $2 trillion and a raise that could reach $100 billion; Nvidia has reportedly considered an anchor investment of up to $10 billion. None of these terms is final. Not everyone is convinced: research firm New Constructs called it the "most ridiculous IPO of 2026" and puts fair value at about $150 billion. Whatever the price, the prospectus will give enterprise buyers something they have never had: audited numbers on Anthropic's margins, customer concentration and compute commitments.

OpenAI: the biggest round ever, and a revenue debate

On 31 March 2026 OpenAI closed a $122 billion round at an $852 billion post-money valuation, co-led by SoftBank with Andreessen Horowitz, D.E. Shaw Ventures, MGX, TPG and T. Rowe Price-advised accounts, and with Amazon, Nvidia and Microsoft participating. It held that valuation through a $7 billion employee tender in August. In late September Bloomberg and Reuters reported preliminary talks to raise at least $30 billion more at roughly $1.4 trillion pre-money.

Revenue is harder to read. OpenAI booked about $13.1 billion in 2025, according to reported figures. In early October Reuters reported that its September run rate was close to $50 billion, about $20 billion less than a number that had circulated as a like-for-like comparison with Anthropic; the gap is largely revenue through cloud partners, which OpenAI does not book. The company has told investors it expects a run rate of $70 billion or more by the end of 2026, driven by enterprise sales. The cost side is just as large: the Financial Times reported projections of about $278 billion in cumulative negative free cash flow from 2026 to 2030 as OpenAI pays for compute. Sam Altman has said a 2026 IPO would be ill-advised; most reports now point to 2027.

Where the money goes: the largest rounds of 2026

Three rounds by two companies, OpenAI's $122 billion and Anthropic's $30 billion Series G and $65 billion Series H, add up to more than all global venture funding in a typical quarter of 2024. Crunchbase counted a record 27 rounds of $1 billion or more in Q3 2026 alone, with Databricks and Safe Superintelligence ($5 billion each) the largest of the quarter.

Mistral: Europe's champion doubles its value

On 8 September 2026 Mistral AI announced a €3 billion Series D at a post-money valuation of more than €21 billion (about $24 billion), the largest equity round ever raised by a European tech company. Samsung Electronics led it, with the EQT-managed Scaleup Europe Fund and existing investor PSG Equity as co-leads. A year earlier, ASML had led the €1.7 billion Series C at €11.7 billion and became Mistral's largest shareholder. In March 2026 Mistral also took $830 million in debt to buy 13,800 Nvidia chips for its own data centre near Paris.

CEO Arthur Mensch says the new money goes into compute and larger models; the recently previewed Mistral Large 4 is a one-trillion-parameter open-weight model. Revenue is still small next to the US labs: analysts at Sacra estimated annual recurring revenue of about $400 million in January 2026, around 60% of it from Europe, and the company targets more than $1 billion this year. For European buyers, the round matters because it makes a sovereign, EU-hosted alternative to the US labs financially credible for years to come.

Manus: independent again, with $500 million

Manus had the most turbulent year of the four. Meta agreed to buy the agent startup for about $2 billion at the end of 2025, but Chinese authorities forced the deal to be unwound. On 8 October 2026 its parent company Butterfly Effect announced a round of more than $500 million, led by Boyu Capital and IDG Capital with Tencent, HSG and ZhenFund following on. The company did not disclose a valuation; Bloomberg had reported beforehand that the round would double it to about $4 billion. The money goes into R&D, GPU capacity and hiring, after the launch of Manus 2.0. For customers, the question that hung over Manus for most of 2026, who will own it, now has an answer: it stays independent.

The venture market: AI takes two-thirds of everything

Global venture funding reached a record of about $300 billion in Q1 2026, driven by OpenAI's round, then $212 billion in Q2 and $159 billion in Q3, according to Crunchbase. Q3 was down 25% on Q2 but up 53% on a year earlier, and the first three quarters together ($679 billion) are the highest ever. AI companies took $102 billion, or 64%, of Q3 funding, 14 percentage points more than a year before. US companies raised 57% of the global total.

Are these valuations justified?

One simple test is to divide the valuation by the latest reported run rate. Anthropic was priced at about 20 times its run rate when the Series H closed and about 15 times on the July figure. OpenAI's $852 billion is about 17 times its September run rate. Mistral and Manus trade at much higher multiples on revenue estimates that are already months old. Fast-growing software companies have rarely been valued above 20–30 times revenue for long, so the US labs are priced for growth to continue, while the smaller players are priced on strategic value: sovereignty in Mistral's case, agent technology in Manus's.

The bear case is the cost side. Both leading labs spend far more on compute than they earn: Anthropic alone has committed $30 billion to Microsoft Azure capacity on top of TPU deals with Google worth tens of billions, and OpenAI's partners are financing data centres worth hundreds of billions. If revenue growth slows before inference becomes cheaper, prices for customers are the most obvious lever. Our analysis of AI inference economics explains why per-token prices keep falling at the low end while premium models stay expensive.

What this means for companies in Croatia and the DACH region

• Vendor survival is not the risk, pricing is. Anthropic and OpenAI have enough capital for years. The real risk for a mid-sized company is that prices, rate limits and plan terms change once the labs answer to public shareholders. Negotiate price protection for 12–24 months and clear notice periods.
• Read the prospectuses. Anthropic's S-1 will be the first audited view of a frontier lab's margins and customer concentration. Use it, and later OpenAI's, in your vendor selection and risk reviews.
• Build for more than one model. A thin abstraction layer and model routing let you move workloads between Claude, GPT and Mistral when prices or terms shift. It is cheap insurance.
• Take the European option seriously. With €3 billion in fresh capital and its own data centre, Mistral is now a long-term option for workloads that must stay in the EU. See our guide to sovereign AI and data residency.
• Check smaller vendors more carefully. Startups that resell or wrap Claude or GPT depend on the labs' pricing and can be squeezed or acquired. Ask about funding runway, data export and exit terms.
• Use the subsidy while it lasts. Today's prices are set by companies that lose money to win market share. Projects with clear ROI that you start now lock in favourable economics and experience.

The bottom line

In 2026 the AI industry became the largest destination for private capital in history. Anthropic went from $9 billion to $65 billion in run-rate revenue and towards a $2 trillion listing; OpenAI raised $122 billion and is talking about $1.4 trillion; Mistral doubled its value to €21 billion; Manus raised $500 million as an independent company again. Behind the headlines are real revenues, real losses and very large compute bills. For companies that use these models, the right response is neither awe nor fear: pick vendors on fit, protect yourself contractually, keep your architecture portable and put the current price war to work.