Pinterest is making its biggest infrastructure move yet. A $4 billion AWS deal aims to transform visual search for over 600 million users. The partnership promises faster AI innovation and sharper shopping experiences. This could reshape how publishers and creators reach audiences.
For anyone building audiences or monetizing content, Pinterest’s latest move signals a seismic shift in how visual discovery will work online. The platform has locked in a $4 billion commitment to Amazon Web Services through 2031, aiming to overhaul its AI-powered visual search for more than 600 million monthly users. This is Pinterest’s largest infrastructure investment to date, and it’s designed to accelerate the rollout of smarter, more responsive search and shopping tools—raising the stakes for publishers, brands, and creators who rely on the platform for reach and revenue.
Pinterest’s leadership sees AI as the key to making discovery more personal and actionable. By deepening its partnership with AWS, the company gains access to advanced compute resources and custom silicon, including Trainium and Graviton chips. These technologies will underpin the next generation of Pinterest’s AI models, powering everything from large language models to vision-language systems that drive personalised recommendations and ad targeting. The deal is structured to support rapid scaling as Pinterest’s ambitions—and user base—grow.
Central to this transformation is Pinterest’s Taste Graph, a sophisticated AI engine trained on billions of images, videos, and text snippets. The Taste Graph not only fuels organic content discovery but also powers ad products like Performance+, which has already shown double-digit improvements in cost per acquisition and slashed campaign setup times for advertisers. By early 2026, it was responsible for nearly a third of Pinterest’s lower-funnel revenue, and its influence is set to expand further as new features roll out.
With the expanded AWS agreement, Pinterest plans to run even more of its infrastructure on Amazon’s custom chips. Graviton processors already handle about a third of Pinterest’s compute needs, and their role will grow as the company pushes for greater efficiency and flexibility. For creators and publishers, these backend upgrades mean faster, more relevant content surfacing—and potentially, new ways to engage audiences and drive conversions as AI-driven discovery becomes the norm.
Graviton, Amazon’s in-house ARM-based CPU family, is engineered for optimal price performance and energy efficiency in the cloud. Its adoption by major platforms like Pinterest highlights a broader industry trend toward custom silicon, as companies seek to balance scale, speed, and cost in the race to deliver next-generation AI experiences.
