Hey hackers! AI became a thing just 3-4 years ago, and by now we see startups pilled with cash from bigtech giants making 100mln ARR in less than a year with new cool models and application. In this article I tried to summarize data from open source to make a list of the biggest AI companies as by now (December 2025). What’s the methodology behind this? What’s the methodology behind this? This ranking includes only AI-first companies—organizations whose core business model is developing and monetizing their own foundation models (LLMs, multimodal models, code models, image/video generators, or autonomous AI systems). core business model To be included, a company must: To be included, a company must: Sell direct access to its models through subscriptions or API usage
Generate the majority of its revenue from those models
Have a publicly reported or credibly estimated annual recurring revenue (ARR) above $100M as of the second half of 2025 Sell direct access to its models through subscriptions or API usage Generate the majority of its revenue from those models from those models Have a publicly reported or credibly estimated annual recurring revenue (ARR) above $100M as of the second half of 2025 annual recurring revenue (ARR) above $100M I intentionally excluded: I intentionally excluded: Big Tech companies like Google, Meta, Microsoft, Amazon, and Apple, because AI is not their primary revenue driver, even though they produce world-class models.
I also excluded infrastructure providers, consulting firms, and startups that rely solely on open-source models without monetizing their own. ARR data comes from verified reports, executive interviews, investor briefings, and reputable financial journalism—including Reuters, The Information, TechCrunch, and VC research platforms. Big Tech companies like Google, Meta, Microsoft, Amazon, and Apple, because AI is not their primary revenue driver, even though they produce world-class models. Big Tech companies not I also excluded infrastructure providers, consulting firms, and startups that rely solely on open-source models without monetizing their own. ARR data comes from verified reports, executive interviews, investor briefings, and reputable financial journalism—including Reuters, The Information, TechCrunch, and VC research platforms. Below is a breakdown of the 11 biggest AI-first companies in the world, ranked by their most recent ARR (second half of 2025). Numbers are based on public statements, financial reports, and reputable media analysis. 11 biggest AI-first companies in the world 1. OpenAI — $13–20 Billion ARR 1. OpenAI — $13–20 Billion ARR OpenAI is the unchallenged leader of the AI-first economy.Fueled by ChatGPT, GPT-4.1, enterprise contracts, and a massive API ecosystem, the company scaled from $1B ARR to over$13B in roughly two years—and is now approaching $20B. $13B $20B With more than 800 million weekly active users, OpenAI has the largest distribution channel in the history of software. Its enterprise suite, now used by over a million businesses, is the engine behind the explosive revenue growth. 800 million weekly active users 2. Anthropic — $7–9 Billion ARR 2. Anthropic — $7–9 Billion ARR Anthropic, creator of Claude, is the fastest-growing enterprise AI provider ever recorded.By late 2025, its ARR jumped to$7B, with internal forecasts nearing $9B. $7B $9B Claude’s reputation for reliability and long-context reasoning made it the preferred option in finance, legal, and security sectors. The company reports that 80% of revenue comes from enterprise API deals—illustrating how deeply AI has penetrated core business operations. 80% of revenue 3. Midjourney — ~$500 Million ARR 3. Midjourney — ~$500 Million ARR The most profitable creative AI company in the world.Midjourney built a $500M+ ARR business withno venture capital, a team of fewer than 50 people, and a fanatically loyal creative community. no venture capital Their image-generation models (V6, V7) set the quality bar for the entire industry. With subscription-only revenue and zero enterprise sales, Midjourney remains one of the purest product-market-fit success stories in modern tech. 4. Cursor (Anysphere) — $500+ Million ARR 4. Cursor (Anysphere) — $500+ Million ARR Cursor is the breakout star of the AI coding revolution (“vibe-coding”).In 2025 it became the fastest-growing developer tool ever, passing$500M ARR and securing a valuation near $30B. $500M ARR $30B Cursor combines a lightweight IDE with a powerful AI pair programmer built on frontier models, enabling developers to modify multi-file projects, refactor code, and even generate full systems. 5. xAI — ~$500 Million ARR 5. xAI — ~$500 Million ARR Elon Musk’s xAI went from zero to half-a-billion ARR in under a year.Its model Grok, integrated deeply into X (Twitter), created a massive built-in distribution channel, while government and enterprise contracts added predictable revenue streams. xAI’s fundraising—over $10B in 2025—ensures it will remain a major contender in frontier model development. 6. Replit — ~$150 Million ARR 6. Replit — ~$150 Million ARR Replit transformed from an online coding sandbox into a fully AI-powered app creation platform.In less than a year, the company jumped from ~$3M ARR to$150M+, driven by its AI agent that can build, run, and deploy applications with minimal human intervention. $150M+ Replit democratizes software creation, bringing coding capabilities to millions who never wrote code before. 7. Hugging Face — ~$130 Million ARR 7. Hugging Face — ~$130 Million ARR The open-source powerhouse of the AI ecosystem.Hugging Face does not rely on a single model; instead, it monetizesinfrastructure, enterprise hosting, private model hubs, model inference, and professional AI tooling. infrastructure, enterprise hosting, private model hubs, model inference With over 100,000 organizations using its platform and ARR exceeding $130M, Hugging Face has become the GitHub of AI. $130M 8. Cohere — $100–150 Million ARR 8. Cohere — $100–150 Million ARR Cohere dominates the enterprise private-LLMsegment.Trusted by banks, governments, and regulated industries, Cohere provides models that run privately—on a company’s own servers or cloud. enterprise private-LLM By focusing on privacy, security, and sovereignty rather than mass adoption, Cohere reached $100M ARR by mid-2025 and now pushes toward the $150M mark. $100M ARR 9. Lovable — ~$100 Million ARR 9. Lovable — ~$100 Million ARR Lovable became a phenomenon in the “vibe coding” wave.In under a year, the company grew from zero to$100M ARR, turning app creation into a creative, remixable experience. $100M ARR Beyond writing code, Lovable generates full application structures—frontend, backend, and deployment—making it one of the fastest-scaling SaaS products in AI history. 10. AI21 Labs — ~$100 Million ARR 10. AI21 Labs — ~$100 Million ARR AI21 Labs specializes in long-context reasoning and enterprise-grade text models.Known for its Jamba and Jurassic families of LLMs, the company secured major partnerships with NVIDIA and Google, pushing revenue toward the$100M mark. $100M Its orchestration engine, Maestro, positions AI21 as a precision-focused alternative to giants like OpenAI and Anthropic. 11. Stability AI — < $100 Million ARR (but influential) 11. Stability AI — < $100 Million ARR (but influential) Stability AI, creator of Stable Diffusion, is the only company on this list not above $100M ARR—yet its impact on the generative ecosystem remains unmatched. not Despite a challenging transition from open-source diffusion models to enterprise monetization, the company continues to innovate across image, audio, and video generation. Follow me — check my profile for links to get more news and insights about AI and how it actually works!

This story contains new, firsthand information uncovered by the writer.

This story contains AI-generated text. The author has used AI either for research, to generate outlines, or write the text itself. 

The 11 Largest AI Companies in the World (December 2025 Edition)

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

AI Metrics Made Simple: Precision, Recall, F-Score, and ROC-AUC

25 Things to Leave Behind in 2025 for Investor Decks

From Correlation to Checkerboards: Model-Free Exploratory Data Analysis for Categorical Data Lands

No Mouse 30: A Quiet Experiment in Keyboard-First Work

How To Find Low-Hanging Fruit Keywords With Strategy

AI Is More Expensive Than You Think

AI Metrics Made Simple: Precision, Recall, F-Score, and ROC-AUC

25 Things to Leave Behind in 2025 for Investor Decks

From Correlation to Checkerboards: Model-Free Exploratory Data Analysis for Categorical Data Lands

No Mouse 30: A Quiet Experiment in Keyboard-First Work

How To Find Low-Hanging Fruit Keywords With Strategy

AI Is More Expensive Than You Think

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps