Pinterest Works with AWS to Power Next Chapter of AI-Driven Visual Search Discovery

Key Terms

data lakes technical

A data lake is a large, centralized digital storage that collects and keeps raw information from many sources—like customer records, market feeds, sensor logs and documents—in their original form. For investors, it matters because a well-built data lake lets a company analyze varied information faster and cheaper, uncovering trends, spotting risks or improving products much like having a single, well-organized warehouse of materials speeds up manufacturing decisions.

transformer-based generative models technical

A class of artificial intelligence systems that generate text, images, code or other content by predicting what comes next, built on a design called a transformer that helps the system focus on the most relevant parts of the input. Think of it as a very large, learned "autocomplete" or recipe book that can combine pieces in new ways. Investors care because these models can create new products, cut labor costs, reshape competitive dynamics, and raise regulatory and ethical risks that affect company value.

multimodal models technical

Multimodal models are AI systems that can understand and generate more than one type of information—such as text, images, audio and video—so a single tool can read documents, interpret pictures and listen to speech. They matter to investors because they can power new products, boost automation and efficiency, shift cost structures and competitive positions, and create technical or regulatory risks; think of them as one employee who can quickly handle tasks that used to require many specialists.

large language models technical

Large language models are advanced AI systems trained on vast amounts of text to understand and generate human-like writing, like a very fast reader and writer that learns patterns in words and sentences. They matter to investors because they can change how companies operate—automating customer service, speeding analysis, cutting costs, creating new products—and they introduce risks around accuracy, security and regulation that can affect a firm’s revenue and reputation.

vision-language models technical

Vision-language models are artificial intelligence systems that connect images and text, allowing a computer to describe pictures, answer questions about images, or generate captions much like a bilingual translator converts between two languages. They matter to investors because they enable new products and efficiencies—such as automated image analysis, smarter search, or customer-facing tools—affecting revenue opportunities, development costs, and competitive advantage, while also introducing data, privacy, and regulatory risks.

kubernetes-based architecture technical

An architecture built on Kubernetes uses an open-source system that organizes and runs software in small, portable pieces across many servers, like a traffic controller directing cars to avoid jams. For investors, it signals a company can scale capacity, deploy updates faster and recover from outages more easily, which can lower operating costs and reduce the risk of revenue loss from downtime.

amazon elastic kubernetes service (eks) technical

A cloud service that lets companies run and manage software packaged in lightweight “containers,” with the platform handling tasks like starting, stopping and scaling those containers automatically. Think of it as a managed port and traffic controller for shipping containers of code: it simplifies operations, speeds deployments and can lower downtime and engineering costs, so investors watch it as a driver of customer growth, recurring revenue and margins for cloud and software businesses.

open-source models technical

Open-source models are artificial intelligence systems whose underlying code, model designs and often training data details are made publicly available so anyone can inspect, modify and use them. For investors this matters because open sharing can lower development costs and accelerate innovation like a shared recipe, create new commercial services and partnerships, but also changes competitive dynamics and introduces risks around security, compliance and how companies can monetize their work.

See more from StockTitan in Google Search and AI answers. Adds StockTitan as a preferred source · opens Google

Add on Google Not now

$4 billion deal, the largest in Pinterest's history, deepens an over decade-long collaboration through 2031

Pinterest expands use of compute, cloud-native architecture, and Amazon custom silicon, including Graviton and Trainium, to power AI at scale

SAN FRANCISCO--(BUSINESS WIRE)-- Pinterest, Inc. (NYSE: PINS) announced a major expansion of its collaboration with Amazon Web Services (AWS), an Amazon.com, Inc. company (NASDAQ: AMZN), its Preferred Cloud Services Provider, including a planned $4 billion commitment for cloud services through 2031. The agreement is the largest infrastructure commitment in Pinterest’s history and is expected to accelerate the company’s AI roadmap, provide a more responsive search and shopping experience, and further modernize the infrastructure powering Pinterest's global visual search discovery platform.

Pinterest Works with AWS to Power Next Chapter of AI-Driven Visual Search Discovery

"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," said Matt Madrigal, Chief Technology Officer, 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. 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.”

A Partnership Built for Scale

Pinterest and AWS have worked together since 2010 to improve the reliability, efficiency and performance of Pinterest's core services, including jointly optimizing one of the largest-scale data lakes on AWS. This renewed agreement significantly deepens that long-standing relationship and is structured to support Pinterest's next phase of growth across AI model training, inference, and platform infrastructure.

"Pinterest is building some of the most advanced visual AI systems on AWS, powering discovery for more than 600 million users. As one of our longest-standing customers, we know what it takes to support that scale securely and efficiently," said Dave Brown, SVP, Compute & ML Services, AWS. "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. This commitment provides Pinterest the AI infrastructure to move faster and deliver new experiences to users sooner."

Scaling AI for Visual Discovery

Pinterest has long applied AI to visual discovery and personalization, and in recent years has accelerated this work with major advances in its recommendation systems and multimodal models. Powered by its proprietary Taste Graph, Pinterest helps users move from open inspiration to personalized, actionable results. The company has evolved from traditional embedding-based retrieval to transformer-based generative models, while continually adapting open-source AI and enhancing its proprietary vision models. Most recently, Pinterest launched Pinterest Assistant, bringing multi-turn conversational discovery to its visual search and discovery experience, powered by open-source vision-language models optimized for scale.

Hardware Optionality and Accelerated Compute

As part of the expanded AWS agreement, Pinterest plans to diversify its use of accelerated compute to support its growing AI needs while improving price performance – and turning to Amazon custom silicon to do it. This includes leveraging AWS Trainium to host and run large language models and vision-language models that power experiences like personalized visual search and AI-assisted discovery. In addition, Pinterest plans to expand its use of Graviton, which already powers roughly a third of the company’s compute infrastructure, to run more of the systems that support discovery for more than 600 million people every month. Together, these investments are expected to give Pinterest greater flexibility to match infrastructure to evolving AI needs.

Modernizing Pinterest's Compute Platform

Pinterest will also continue a major infrastructure modernization effort under the agreement, transitioning from traditional EC2-based environments to a Kubernetes-based architecture on Amazon Elastic Kubernetes Service (EKS). The migration is expected to improve developer velocity, operational reliability and infrastructure efficiency across Pinterest's global platform.

These investments are intended to strengthen and refine Pinterest’s AI infrastructure foundation and support its continued innovation in AI-powered visual search and discovery.

About Pinterest

Pinterest is a visual search and discovery platform where people find inspiration, curate ideas and shop products — all in a positive place online. Headquartered in San Francisco, Pinterest has over 600 million monthly active users worldwide.

View source version on businesswire.com: https://www.businesswire.com/news/home/20260604200750/en/

Press:
press@pinterest.com

Investor relations:
ir@pinterest.com

Source: Pinterest, Inc.