Pinterest has expanded its partnership with Nvidia to improve AI-powered discovery. The new foundation combines Nvidia Blackwell GPUs, Nvidia Dynamo and Pinterest’s own visual embeddings. According to Pinterest, it gives teams “a shared way to build and launch multimodal AI products without creating custom infrastructure for each one” — more AI features with less infrastructure investment.
The deal extends an existing agreement and sets up a long-term platform relationship for the next generation of visual and conversational experiences. It will power the next stage of Pinterest AI visual search through expanded content understanding and improved safety systems.
The result is more personalized, responsive AI search — and more of Pinterest’s user data put to work on performance.
Discovery on Pinterest starts with images, not text. The better the system understands images, the more product image quality, consistency and catalog accuracy determine visibility. A range of styled shots, in-use scenes and detail images gives an Assistant that uses more visual context more to match.
Turning searches into shopping signals connects directly to ad products such as Pinterest’s visual search ads and AI restyle. Faster responses encourage more exploration, and more exploration creates more ad inventory.
Using more user data sharpens personalization but also makes it harder for brands to predict what the platform will surface. Break reporting down by category and image type to learn which creative the AI recommends. For the wider ad roadmap, see Pinterest Presents’ ad updates.
It creates a shared AI foundation on Nvidia Blackwell GPUs, Nvidia Dynamo and Pinterest’s visual embeddings, improving visual search, content understanding and safety systems.
Pinterest Assistant can use 25x more visual context per request, and benchmarks showed about 85x faster response startup and 7.3x lower overall latency.
Invest in image quality and variety and accurate catalog data, and analyze performance by image type to learn what the AI surfaces.
To apply what you just read to your own site, start with a free audit of where things are now.
A strategist replies within 24 hours on business days.
TikTok’s Road to Black Friday series shows how top TikTok Shop sellers prep for BFCM: Princess Polly’s 6,000 creator videos a month, Dossier’s sampling engine and SharkNinja’s LIVE strategy.
Meta shared how small businesses use its Muse AI agents: analyzing ad performance, building campaigns through conversation and automating workflows with Shopify, Canva and HighLevel. Prompts included.
Reddit outlined its U.S. midterm election plans, including the Up the Vote hub, AI detection that cut spam exposure 20% and removes nearly 2 million fake upvotes daily, and AMA rules for political ads.