Anthropic: from 1 to 4 billion in annualized revenue in 7 months
Anthropic, the company that created Claude, is posting unprecedented growth: its annualized revenue rose from 1 billion dollars in December 2024 to more than 4 billion in July 2025. Valued at 61.5 billion at its last funding round in March 2025 and on track for an even higher valuation, Anthropic is now one of the fastest-growing AI companies in the world. A strong signal of how quickly businesses are adopting AI.
Opportunities for businesses
Beyond the headlines about valuations, this funding round has very concrete consequences for the SMBs and mid-market companies that use or are considering using AI. Three positive dynamics are taking shape.
Continued price drops
The fierce competition between Anthropic, OpenAI and Google is pushing rates down. The cost of a Claude API request fell by 75% between 2024 and 2025. For an SMB that consumes 50 million tokens per month, the bill goes from 750 € to less than 200 €. This trend will continue as long as the race for market share lasts.
Faster innovation
Billions in revenue allow Anthropic to invest massively in R&D. Models improve every 3 to 6 months: longer contexts, more reliable reasoning, agentic capabilities. Each new generation solves problems that were impossible 6 months earlier.
Strengthened enterprise offerings
Investors demand a return: AI providers are building serious enterprise offerings with SLAs, dedicated hosting, GDPR compliance and premium support. Anthropic launched its Teams and Enterprise offering in 2025, making Claude viable for critical enterprise use cases.
Risks and points of caution
This concentration of capital in a handful of American players raises legitimate questions for European companies.
Risk of technological dependence
With three American players (OpenAI, Anthropic, Google) dominating the LLM market, European companies are exposed to a risk of dependence. A change in pricing policy, a geographic restriction or a modification of the terms of use could directly impact your operations. Mistral AI, a French player, remains a credible alternative but with resources 20 times smaller.
Data sovereignty
The massive valuations of AI players rest in part on the value of the data processed. Systematically check the confidentiality and data-use clauses in your API contracts. Favor deployments via European regions (available from Anthropic through AWS eu-west) and enable data non-retention options.
Our recommendations
Faced with an accelerating market, companies must structure their AI approach to take advantage of falling prices while limiting the risks of dependence.
Adopt a multi-provider strategy
Don't bet everything on a single model. Use an abstraction layer (LiteLLM, LangChain) so you can switch between Claude, GPT and Gemini without rewriting your code. Test each new model on your specific use cases: the best overall model is not necessarily the best for your need.
Take advantage of the price war now
Prices have never been so low and the downward trend continues. It is the ideal time to launch AI pilot projects: the cost of experimentation is minimal. Identify 2 to 3 high-value use cases and launch POCs with a budget of 500 to 2,000 € over 4 weeks.
Put Mistral AI on your radar
The French player Mistral AI offers competitive models with native European hosting. If data sovereignty is a stake for your sector (healthcare, defense, finance), test Mistral Large alongside Claude and GPT. Diversification strengthens your negotiating position.
Key takeaways
Frequently asked questions
Why does Anthropic attract so many investors?
Anthropic stands out for its "safety-first" approach to AI, its Claude model recognized as one of the best performing on the market, and its annualized revenue growth from 100 million to more than 2 billion dollars in 18 months. Investors are betting on its ability to capture the enterprise market thanks to superior reliability.
Will this growth drive up API prices?
On the contrary. The race for investment pushes players to lower their prices to gain market share. API rates have dropped by 80% in 18 months. This trend should continue at least until 2026, making AI ever more accessible to SMBs.
Should we worry about dependence on a single AI provider?
It is a real risk. The best strategy is to design your AI applications with an abstraction layer (LiteLLM, LangChain) that lets you switch between providers. Test the alternatives regularly and avoid using proprietary features that cannot be reproduced elsewhere.
Major players in the LLM market
Valuation: $100B
Model renowned for reliability and safety. Claude 3.5 Sonnet dominates reasoning and coding benchmarks. "Constitutional AI" approach for safer responses. Fastest growth in the sector in 2025.
Valuation: $150B
The most widely used model in the world. A complete ecosystem with ChatGPT, API and Microsoft integration. Dominant position in the consumer space and early enterprise deployments. The most versatile model available.
Valuation: €6B
France's AI champion. High-performing models with native European hosting. A credible alternative for sectors sensitive to data sovereignty. Backed by Microsoft and the French government.
API pricing (per million tokens)
Comparison
| Criterion | Claude (Anthropic) | GPT-4o (OpenAI) | Mistral Large |
|---|---|---|---|
| Reasoning | ⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐ |
| Code | ⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐ |
| Enterprise offering | ⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐ |
| EU sovereignty | ⭐⭐ | ⭐ | ⭐⭐⭐ |
| Value for money | ⭐⭐⭐ | ⭐⭐ | ⭐⭐⭐ |