{"title":"Anthropic vs Safe Superintelligence: Frontier labs as compute buyers compared","url":"https://www.frontiertechintelligence.com/compare/anthropic-vs-safe-superintelligence","comparison":"Anthropic ranks 59 on the AI Infra 100 with a score of 89; Safe Superintelligence ranks 98 with 69. Anthropic is ahead on deployment, production, commercials and positioning. Safe Superintelligence is ahead on progress. Level on technology, firepower and leverage. Anthropic is based in San Francisco, United States, Safe Superintelligence in Palo Alto, United States. Anthropic: private. Anthropic is named in 3179 signals and Safe Superintelligence in 6.","companies":[{"name":"Anthropic","url":"https://www.frontiertechintelligence.com/companies/anthropic","city":"San Francisco","country":"United States","stage":"private","founded":null,"last_round":null,"positioning":"Anthropic is a U.S. public benefit corporation developing Claude and operating as a frontier-model compute buyer rather than an owner-operator of large data-center sites. It signed agreements in 2025-26 for up to 5 GW of Amazon Trainium capacity, about 5 GW of next-generation Google/Broadcom TPU capacity, and GPU capacity from SpaceX, while making Claude available through AWS, Google Cloud, and Microsoft Azure. Its distinguishing position is diversified access to multiple hyperscaler and neocloud platforms; the open question is how much of its announced capacity is energized and operational rather than contracted.","assessed_in":{"domain":"AI Core","list":"AI Infra 100","rank":59,"score":89,"tier":"member","scores":{"leverage":10,"position":10,"progress":9,"firepower":10,"commercial":10,"deployment":10,"production":8,"technology":6},"why":"Anthropic develops and sells Claude, a closed-weight frontier AI model and assistant for enterprises, developers, and consumers.","segment":"Frontier labs as compute buyers","layer":"Compute demand & operators"},"signals_total":3179,"facts_total":492},{"name":"Safe Superintelligence","url":"https://www.frontiertechintelligence.com/companies/safe-superintelligence","city":"Palo Alto","country":"United States","stage":null,"founded":null,"last_round":null,"positioning":"Research lab with no product; compute comes through a Nvidia partnership (Vera Rubin access) and Nvidia equity.","assessed_in":{"domain":"AI Core","list":"AI Infra 100","rank":98,"score":69,"tier":"member","scores":{"gate_ne":["technology","firepower"],"leverage":10,"position":6,"progress":10,"firepower":10,"commercial":2,"deployment":4,"production":6,"technology":6},"why":"Safe Superintelligence develops methods for controlling rogue AGI, directly qualifying as AI safety and alignment work.","segment":"Frontier labs as compute buyers","layer":"Compute demand & operators"},"signals_total":6,"facts_total":6}],"note":"Assessment by Frontier Tech Intelligence; facts and signals carry their sources."}