The research AI startup anthropic aims to raise up to $5 billion over the next two years to take on rival OpenAI and enter more than a dozen major industries, according to company filings TechCrunch received.
A pitch deck for Anthropic's Series C fundraising round reveals these and other long-term goals for the company, which was founded in 2020 by former OpenAI researchers. On deck, Anthropic says it plans to build a "marginal model" — which it tentatively calls "Claude-Next" — 10 times more capable than the most powerful current AI, but that this will require spending a billion dollars over the next 18 months. .
Anthropic describes the boundary model as a "next-generation algorithm for self-teaching AI," referencing an AI training technique it developed called "constitutional AI."
At a high level, constitutional AI seeks to provide a way to align AI with human intentions – allowing systems to answer questions and perform tasks using a simple set of guiding principles.
Anthropic estimates that its frontier model will require 10^25 FLOPs, or floating point operations – a size far greater than even today's largest models. Of course, how this translates into computational time depends on the speed and scale of the system doing the computation. Anthropic implies (in the deck) that it will be based on clusters with “tens of thousands of GPUs.”
