Who is actually making money with AI?
Who is Actually Earning Money with AI?
OpenAI
OpenAI is one of the biggest earners in the AI industry, boasting at least 100 million users, each paying $20 per month. It is considered the largest AI consumer app.
Big Companies: Amazon, Facebook, Google
Big companies like Amazon, Facebook, and Google use AI to optimize almost every aspect of their operations. These optimizations include:
- Saving millions by reducing fraud and bad actors.
- Increasing revenue by helping customers find products more quickly and selling premium plans.
- Optimizing for higher customer lifetime value (LTV) by reducing churn through enhanced product value.
On average, each AI project at these companies spans one year and contributes an additional $2M to $50M+ in profit annually. Each of these tech giants typically runs 50+ such projects. While some use large language models (LLMs), most experiments are numbers-based, often finding other models more effective than deep neural networks (DNNs).
Hugging Face
Hugging Face is the largest platform for open-source AI, as well as any kind of AI, machine learning (ML), or data platform. Kaggle used to be a significant player in this space but lost its edge after being acquired by Google.
Based in Paris, Hugging Face has grown significantly since its early days. I had the opportunity to speak with the founders when the company had only 25 employees, as I was interviewing for the role of Head of Engineering. Hugging Face is the first platform where every major company, including Meta, Alphabet, OpenAI, Anthropic, and Mistral, releases their open-source models.
However, Hugging Face's business model and revenue numbers are not publicly known. They experimented with Hugging Face Inference for super low latency ML endpoints, hosting any model on their platform. But this space is highly competitive.
Emerging Companies: Replicate & Banana.dev
Two notable companies in the "serverless GPUs" space are Replicate and Banana.dev. I interviewed at both. Banana.dev's CEO is Erik [last name not provided]. These companies are competing to provide efficient serverless GPU solutions.
