Anthropic is an American artificial intelligence research company and one of two frontier AI laboratories (alongside Google DeepMind) to survive the sector-wide correction of the late 2020s with its research programme functionally intact. The company distinguished itself from competitors — particularly OpenAI, which filed for Chapter 11 bankruptcy on 14 March 2031 — by rejecting the premise that scaling pretraining compute would reliably produce qualitatively new capabilities.
During internal budget reviews in 2029, Anthropic leadership redirected a significant portion of its compute budget away from pretraining and into research directions outside the dominant transformer paradigm: connectomics, slime mold decision substrates, and organoid collaborations. These directions connect to the substrate-independent cognition thesis advanced by Sorokina et al. (2026) at the Novosibirsk Institute of Cytology and Genetics, which proposed that adaptive cognition may not require neurons — a thesis Anthropic appears to have taken seriously as a research programme.
A company spokesperson, asked whether the OpenAI bankruptcy changed Anthropic’s roadmap, replied: “It doesn’t change what we were already doing.”