Pinterest teams with Nvidia to better support AI shopping
Pinterest is collaborating with a pioneer in next-gen artificial intelligence to spur its evolution as a digital commerce platform.
The social network has built a new AI layer on the Nvidia accelerated computing solution with Nvidia Blackwell graphics processing units (GPUs) and Dynamo generative AI framework and its own visual embeddings. Pinterest seeks to utilize the foundation as a shared way to build and launch multimodal AI products without creating custom infrastructure for each one.
Pinterest intends to turn its more than 80 billion monthly searches into signals for AI-based discovery and shopping, enabling its Pinterest Assistant visually focused artificial intelligence tool that aids online shopping and discovery to run 25 times more visual context per request.
[READ MORE: Pinterest pilots ‘visual-first’ AI shopping assistant]
In benchmark testing, Pinterest says it saw average improvements of about 85 times faster response startup and 7.3x times faster overall latency when using precomputed visual representations instead of repeatedly processing raw images.
Pinterest initially developed its foundational system to support the intelligence layer that runs throughout its user experience. This foundation already supports a broader set of AI use cases across Pinterest’s platform, including ranking, safety, optical character recognition (OCR), signal generation and future experiences.
The expanded partnership is also expected to provide the stronger infrastructure needed to support Pinterest’s growing use of vision-language models (VLMs) faster and more efficiently at scale.
“Building the next generation of AI-powered discovery means investing in infrastructure that can keep up with the scale and complexity of Pinterest,” said Kartik Paramasivam, chief architect at Pinterest. “Our collaboration with Nvidia helps us deliver faster, smarter and more personalized experiences for the hundreds of millions of people who use Pinterest.”
Other AI functionality Pintertest has deployed in the last 18 months includes introducing a set of new visual search features that allows users to further interact with image pins for women’s fashion content, an AI-based feature called “Promote a Pin” that enables any user to boost their content on the platform with a few taps and no advertising experience requirement, and a tool known as “Scent Scanner” which translates visual preferences from customer Pinterest boards into personalized Estee Lauder Jo Malone London fragrance recommendations.
“Pinterest is transforming visual discovery with AI, helping hundreds of millions of people find inspiration through more intelligent and personalized experiences,” said Ujval Kapasi, VP, AI & HPC frameworks and libraries at Nvidia. “By building its multimodal AI infrastructure on Nvidia accelerated computing and software, Pinterest can bring the next generation of visual and conversational AI experiences to its users at scale.”
Pinterest and Nvidia are building on an existing five-year collaboration across more than 14,000 Nvidia GPUs. By standardizing the infrastructure behind multimodal AI, Pinterest hopes to create a long-term platform for the next generation of visual and conversational experiences across its enterprise.
