As Anthropic gains momentum in the booming artificial intelligence market, OpenAI sent a memo to investors this week slamming its main competitor for a "significantly narrow development curve" and saying it has limited computing resources. According to the memo, OpenAI plans to achieve a computing power scale of 30 gigawatts by 2030, while Anthropic is expected to only reach about 7 to 8 gigawatts by the end of 2027.
"Even based on the upper limit of its range, our expansion speed is significantly ahead, and the gap is still widening." OpenAI said.
In a market where large-scale language models and artificial intelligence services are rapidly iterating, the two companies are fiercely competing for industry status, and OpenAI's concerns about Anthropic are also growing. The two companies, with a total valuation of over one trillion US dollars, are preparing for IPOs that may be launched this year. At the same time, they are trying to prove to investors that they have sustainable business models and can withstand the pressure of competing with well-funded companies such as Google and Meta.
Anthropic was co-founded in 2021 by CEO Dario Amodei and a group of researchers and executives who left OpenAI. Earlier this week, Anthropic, which has performed well in the enterprise market, released a powerful new model that will be rolled out to select enterprises as part of its new cybersecurity project called "Project Glass Wings."
In this investor memo, OpenAI quoted Amodudi’s statement that it deliberately adopts a conservative computing power strategy, and said that its own strategy is “different from it.”
“Each generation of new infrastructure allows us to train more capable models, making each token smarter than the previous generation,” OpenAI wrote. “At the same time, algorithm optimization and hardware upgrades will reduce the service cost of a single token, thereby lowering the cost of unit intelligence.”
OpenAI, which kicked off the generative artificial intelligence boom with ChatGPT at the end of 2022, said that the company is enjoying the "compound interest advantage": better infrastructure and models reduce costs, and better products bring higher revenue.
"This advantage also allows OpenAI to continue to promote the popularization of AI: we open our tools to hundreds of millions of people for free, are more generous to developers, and empower technical capabilities to those who use our tools to create and solve problems," the company said.
Anthropic did not immediately respond to a request for comment.