OpenAI co-founder and former Chief Scientist Ilya Sutskever has partnered with Google Cloud to use the company’s TPU chips in his newest artificial intelligence venture, Safe Superintelligence (SSI). According to an announcement from Google Cloud, SSI will leverage these tensor processing units to significantly accelerate its research and development efforts aimed at creating safe and advanced AI systems.
Cloud providers have increasingly targeted high-profile AI startups capable of spending hundreds of millions of dollars annually on computational resources needed for training large-scale foundational models. The partnership notably positions Google Cloud as SSI’s primary computing provider, according to sources familiar with the arrangement.
This collaboration continues Google’s extensive history of partnering strategically with former AI researchers who have gone on to found successful billion-dollar enterprises. Sutskever once worked at Google’s research division, Google Brain, contributing significant developments to neural network technologies. Last year, ex-Google Cloud AI Chief Scientist Fei-Fei Li chose Google Cloud to serve as the main cloud provider for her own startup, World Labs.
It remains unclear if SSI has established additional agreements with other computing providers; neither Google Cloud nor SSI representatives provided further comment on the subject.
SSI publicly emerged from stealth mode in June 2024, several months following Sutskever’s departure from OpenAI, where he was instrumental in pioneering research into AI safety. The startup secured $1 billion in funding from prominent investors such as Andreessen Horowitz, Sequoia Capital, DST Global, and SV Angel.
Since its launch, SSI has revealed few concrete details about its operations, stating only that its primary objective—safely developing superintelligence—is integral to its core mission. Sutskever previously mentioned exploring innovative ways to boost the performance and capabilities of frontier AI architectures.
Before co-founding OpenAI, Sutskever refined his expertise at Google Brain, where he focused extensively on deep learning techniques. He was deeply influential within OpenAI, notably playing an integral role during internal conflicts which saw CEO Sam Altman removed and later reinstated in November 2023. Following those events, Sutskever gradually distanced himself from day-to-day activities at OpenAI, eventually leaving to establish Safe Superintelligence.
