The vast market for language models may be heating up, but the CEO of AI powerhouse Anthropic isn’t really worried about competition, predicting that the price of training such models could eventually reach $100 billions of dollars.
Anthropic CEO and co-founder Dario Amodei discussed the future of Anthropic, its chatbot “Claude” and the billion-dollar AI industry in a wide-ranging interview with CNBC last week.
“The number of players” who have the financial capacity to train enterprise-grade AI models at scale, Amodei said, “will be relatively small initially.”
Amodei and his sister, Daniela Amodei, co-founded Anthropic in 2021, quickly attracting major backers including Amazon. The e-commerce company invested $1.25 billion in Anthropic last year and pledged another $2.75 billion in March, cementing a powerful partnership that gives Amazon a minority stake in Anthropic and enables Anthropic to access Amazon’s servers and cloud chips.
The company’s Claude competes with similar models like OpenAI’s ChatGPT and Google Gemini. Meanwhile, Amazon is working on its own “Olympus” AI, while Elon Musk open sourced his “Grok” model last month.
But even as model development booms, Amodei dismissed concerns about rapid commodification, pointing to the astronomical price of creating and training large language models. Amodei told CNBC that developing current models already costs a company $100 million — and that price will only increase as the technology advances.
“I think training the models next year will cost about $1 billion,” Amodei told the outlet. “And then by 2025, 2026, we’ll hit $5 billion or $10 billion. And I think there’s a chance it could go beyond that to $100 billion.”
The number of companies financially able to train models at this cost will remain small, he said.
Diversity in development techniques can also help avoid commodification, Amodei told CNBC, comparing different models to differences between human beings.
“As humans, we’ve all done it: our brains are basically all designed the same way, but we’re very different from each other, and I think the patterns will be the same,” he said. he declares.
Amodei said some AI models could specialize in topics such as law or national security, while others could gain expertise in biochemistry.
“I think this strength will lead to different model providers specializing in different things, even if the basic model they created is the same,” he added.
businessinsider
