Pinterest commits $4 billion to AWS cloud services through 2031

Pinterest's $4 billion AWS deal through 2031 is the company's largest infrastructure commitment, betting on custom silicon to power AI-driven visual discovery for 600 million users.

Pinterest deepened its decade-long partnership with Amazon Web Services, committing $4 billion for cloud services through 2031 in the largest infrastructure deal in the company's history, as it races to scale AI-powered visual search for more than 600 million monthly active users.

"This expanded commitment with AWS gives us the compute flexibility, hardware optionality, and infrastructure efficiency to accelerate our AI vision for the next generation of visual discovery on Pinterest," Matt Madrigal, chief technology officer at Pinterest, said.

The agreement extends a collaboration dating to 2010 and covers AI model training, inference, and platform infrastructure. Pinterest plans to use AWS Trainium to host large language models and vision-language models powering features like personalized visual search and the recently launched Pinterest Assistant, a conversational discovery tool. The company will also expand its use of AWS Graviton processors, which already power roughly one-third of its compute infrastructure.

For Amazon, the deal locks in significant recurring revenue from a major client through 2031, reinforcing AWS's dominance in cloud infrastructure at a time when hyperscalers are competing fiercely for AI workloads. For Pinterest, the commitment provides the compute capacity needed to serve its global user base while modernizing its infrastructure.

Pinterest has long applied AI to visual discovery and personalization through its proprietary Taste Graph, which helps users move from open-ended inspiration to actionable results — finding a recipe, shopping a look, or planning a home renovation. The company has evolved from traditional embedding-based retrieval to transformer-based generative models, and its latest product, Pinterest Assistant, brings multi-turn conversational discovery powered by open-source vision-language models optimized for scale.

AWS Custom Silicon Becomes a Strategic Asset

The deal highlights the growing importance of Amazon's custom silicon strategy. Pinterest will use AWS Trainium to run the large language models and vision-language models that underpin personalized visual search and AI-assisted discovery. AWS Trainium, Amazon's purpose-built AI training chip, competes directly with Nvidia's H100 and B200 GPUs, offering an alternative for companies seeking to diversify their AI hardware supply chains. Pinterest's expanded use of Graviton — which already handles a third of its compute load — further reduces reliance on traditional x86 processors from Intel and AMD.

The agreement also includes a major infrastructure modernization component. Pinterest will migrate from traditional EC2-based environments to a Kubernetes-based architecture on Amazon Elastic Kubernetes Service, a move expected to improve developer velocity, operational reliability, and infrastructure efficiency.

For Amazon, the $4 billion commitment through 2031 adds to a growing roster of long-term cloud deals that provide revenue visibility for AWS, the company's most profitable segment. Amazon generated $742.78 billion in revenue over the last twelve months as of Q1 2026, with AWS contributing a significant share of operating income. Pinterest shares, trading on the New York Stock Exchange under the ticker PINS, fell 0.96% on the day of the announcement, while Amazon shares declined 2.53%.

The deal underscores the intensifying competition among hyperscalers — Microsoft Azure and Google Cloud are also vying for large-scale AI workloads — and positions AWS to capture a larger share of the AI infrastructure buildout. For Pinterest, the investment provides the computational foundation to continue innovating in visual search and discovery, a category where AI capabilities increasingly determine user engagement and advertiser returns.

This article is for informational purposes only and does not constitute investment advice.