Pinterest Commits US$4bn to AWS for Ad Performance

Pinterest has committed US$4bn to AWS through to 2031 as part of an expanded partnership with the cloud services provider. The agreement could accelerate the company's work on machine learning and improve both search performance and advertising effectiveness.

According to Pinterest, the investment is the largest infrastructure commitment in the company's history. The platform uses AI to power visual discovery for more than 600 million users and to support advertiser performance through personalised recommendations.

"Pinterest is heavily investing in AI to make discovery more personal, visual and actionable for the hundreds of millions of people who use our platform every month," says Matt Madrigal, CTO at Pinterest. "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 also noted the advertising benefits. "This strategic partnership will help accelerate AI innovation at Pinterest, improving both our consumer experience and advertiser performance by advancing our proprietary models and our use of open-source models," he says.

Infrastructure for advertiser outcomes

Pinterest has worked with AWS since 2010 to support its core services. The company uses its Taste Graph technology to move users from broad inspiration to what it describes as actionable ideas.

The platform has updated its systems with gen AI and continues to develop its visual AI capabilities. Pinterest uses both proprietary models and open-source alternatives to power discovery and advertising targeting.

The Pinterest Assistant feature uses open-source vision-language models to enable multi-step visual search conversations. According to the company, this technology could improve how users discover products and interact with advertised content.

"Pinterest is building some of the most advanced visual AI systems on AWS, powering discovery for more than 600 million users," says Dave Brown, SVP of Compute & ML Services at AWS. "As one of our longest-standing customers, we know what it takes to support that scale securely and efficiently. AWS compute and purpose-built silicon like Trainium and Graviton give Pinterest the price-performance to train and run AI models at massive scale across both training and inference."

Expanded compute for model training

Pinterest plans to use AWS Trainium to host large language models and vision-language models for personalised visual search. The company says this technology supports its AI-assisted discovery features.

AWS Graviton already powers around a third of Pinterest's compute, according to the firm. Pinterest will expand this to support discovery for its monthly user base.

Matt Garman, CEO of AWS, wrote on LinkedIn about the partnership. "Over the last 16 years, I've watched Pinterest evolve from a visual pinboard into an AI-powered discovery platform serving more than 600 million users around the world," he says. "Most recently, that includes Pinterest Assistant, which helps people explore ideas through natural conversation."

Matt also praised the Pinterest team's focus. "I've always appreciated how focused Bill Ready and this team are on improving the experience for their users, and it's exciting to see that innovation continue to accelerate," he says.