Pinterest announced a historic $4 billion commitment to Amazon Web Services (AWS) to expand its AI infrastructure through 2031, marking the largest infrastructure investment in the company's history. This deal will enable Pinterest to accelerate the development and deployment of AI models that enhance visual discovery and personalization across its platform.
The agreement extends a partnership that began in 2010, during which Pinterest and AWS have jointly optimized one of the largest-scale data lakes on AWS. The new commitment focuses on advancing AI training, inference, and platform infrastructure to support personalized content discovery for more than 600 million monthly users worldwide.
To power this AI-driven growth, Pinterest will utilize AWS Trainium chips to train and run large language and vision-language models that underpin its visual search and recommendation technologies. Additionally, the company plans to broaden its use of AWS Graviton processors, which currently account for nearly a third of Pinterest’s compute infrastructure, to enhance system efficiency and scalability.
Pinterest leverages AI extensively through its proprietary Taste Graph, which connects users from general inspiration to actionable outcomes such as shopping or project planning. Recent AI innovations include transformer-based generative models and a conversational interface called Pinterest Assistant, which offers a multi-turn conversational experience for visual discovery using open-source vision-language models optimized at scale.
This infrastructural upgrade aligns with Pinterest's goal to make discovery more personal, visual, and actionable. The company’s chief technology officer emphasized that this collaboration with AWS provides the compute flexibility and hardware options necessary to accelerate Pinterest’s AI vision for the future of visual discovery.