More than two years after the blockbuster launch of ChatGPT, artificial intelligence continues to be the white hot center of venture capital and the business world at large. Some of the buzziest startups have shifted focus from the AI model release horserace to building useful applications and products on top of existing models, automating scutwork across fields like engineering, healthcare, legal and sales. That’s evident in Forbes’ seventh annual AI 50 list, produced in partnership with Sequoia and Meritech Capital, which spotlights the most promising privately-held AI companies in the world.
Newcomers to the list include $2.5 billion-valued Anysphere (better known as Cursor), a three year-old AI startup that helps engineers write and edit code and has at least $100 million in annualized revenue; the $1 billion-valued AI language tutor app Speak, which is used by some 10 million people to learn English and Spanish; and Massachusetts-based unicorn OpenEvidence, which is building an AI-powered search platform that summarizes medical information for doctors.
Model builders like OpenAI and Anthropic remain the biggest companies on the list, having raised large sums of cash from marquee Silicon Valley venture capitalists. The two AI behemoths have accumulated a combined $81 billion in venture funding, more than half of the total $142.45 billion that companies on this year’s AI 50 list have raised. But this year, there’s also a hefty new competition from Elon Musk’s xAI ($12.1 billion in funding) and former OpenAI CTO Mira Murati’s stealthy startup Thinking Machine Labs, which is developing broadly capable AI systems and reportedly raising $1 billion at a valuation of about $9 billion. Meanwhile, Fei Fei Li, known as the “godmother of AI”, has launched her own startup too, World Labs ($291.5 million raised), to develop models that can understand physical spaces. And then there’s enterprise AI company Writer ($326 million raised), which is training its own models to handle mundane business tasks like writing marketing blogs and searching through troves of documents. (Read our feature story on Writer here.)
Artificial intelligence companies (whether apps or model makers) typically rely on expensive silicon chips and energy-intensive data centers for computing power to train and run their models and applications. That’s been advantageous for a cohort of infrastructure providers like $2.8 billion-valued Crusoe, $2.5 billion-valued Lambda and $3.3 billion-valued Together AI, which have bubbled up to sate AI’s voracious appetite for compute. (See our deep dive on Crusoe.)
But as companies spend billions of dollars on computing power to train ever larger, more capable models, startups like Chinese dark horse DeepSeek have shown that training can be done in a more cost efficient way. While DeepSeek isn’t on the AI 50 list this year because the company’s funding, revenues and business practices are opaque, you’ll find a cohort of Chinese AI companies to watch here.
Some of these startups are grappling with industry-wide issues, one of the biggest being litigation over alleged copyright infringement that is still moving through the courts. Multiple startups on the list, including OpenAI and Anthropic, have been sued for allegedly training their AI models on millions of copyrighted works of art and intellectual property. Meanwhile, a growing group of publishers, artists, musicians and authors have expressed concerns that AI companies have illegally scraped their data from the internet to create tools that then compete with them, posing a threat to their livelihoods. Image and video generation tools like Runway and Midjourney were similarly sued by a group of artists in 2023. Elevenlabs, which develops AI tools to create cloned voices, has faced lawsuits from voice actors. Music generator Suno has faced copyright infringement-related legal action from major record labels for its music generation technology. Cohere and Perplexity have been met with criticism and legal action from a flurry of news publishers for alleged copyright infringement. Forbes sent a cease-and-desist letter to Perplexity in June 2024, accusing it of infringing copyright and joined a class action lawsuit against Cohere in February for similar reasons. AI companies have broadly argued that their use of publicly available copyrighted content falls under fair use. But the future of AI ultimately hinges on court rulings on these suits.
This year was among the most competitive for the AI 50 list. Forbes received some 1,860 submissions. Applicants do not pay a fee to be considered and are judged for their business promise, technical talent and use of AI through a quantitative algorithm and qualitative judging panels. Companies are encouraged to share data on diversity, and our list aims to promote a more equitable startup ecosystem. For more, see our full package of coverage, including a detailed explanation of the list methodology, videos and analyses on trends in AI.
Editorial Director: Elisabeth Brier
Reporters: Richard Nieva, Amanda Florian, Stephen Pastis and Leah Rosenbaum