Most startups and emerging companies won't be able to afford to get into the AI race, according to the CEO of Anthropic, Dario Amodei, the company behind Claude, which is valued at $18 billion and analyzed the evolution of AI on the podcast In Good Company.
https://www.youtube.com/watch?v=xm6jNMSFT7g
In an interview with the CEO of Norges Bank, Nicolai Tangen, Amodei discussed how long it takes to bring an AI model to the market and how much it can cost to introduce this technology. According to Amodei, the cost of training these technologies is around 100 million dollars, but "today there are models in training that cost more than 1 billion dollars."
For the Anthropic CEO, the current prices are just the starting point, and in the next few years they will increase, because "maybe by 2027 we will reach 10 or 100 billion dollars, and then the chances are that we will have models that are better than most people in most things."
High standard
The high price of developing AI makes it difficult for startups to enter the market and develop models that can compete with tech giants like OpenAI or Anthropic. On average, US emerging companies raised around 59 million dollars in funding in the first quarter of 2023, a far cry from the 450 million dollars raised by Anthropic in May 2023.
That's why, Amodei said, small companies will have a hard time entering the AI race: "It's going to be a vibrant ecosystem downstream and a small model ecosystem," the Anthropic executive said, suggesting that small companies won't be able to compete with the big ones.
In the podcast, Amodei also revealed that Anthropic is working on AI chips to compete with Nvidia, with the help of Google and Amazon, which have invested $2 billion and $4 billion respectively in Anthropic.
