Unified data foundations over a massive multi-year initiative
Achieved 99.999% availability for critical AI workloads
Reduced model deployment timelines from weeks to minutes
Empowered 10,000+ engineers to build with agentic AI
With Google Cloud, PayPal built a unified AI platform to architect the next era of commerce for over 439 million global users.
With Google Cloud, PayPal built a unified AI platform to architect the next era of commerce for over 439 million global users.
Laying the foundations for an AI-first world
For over 25 years, PayPal has been revolutionizing commerce, empowering over 439 million consumers and merchants in nearly 200 markets to thrive in the global economy. Now, as online shopping shifts towards an agentic future—where AI can act on a consumer’s behalf—PayPal is once again paving the way for this next wave of innovation.
“Every FinTech player is now racing to build AI-native infrastructure,” says Srinivasan Manoharan, PayPal’s director of AI/ML platform. “It is no more about redefining payments—it is about who can build that platform, which is governed and data-ready.”
True to its pioneering spirit, PayPal has already made significant investments to modernize its technology stack and unify its data infrastructure. After completing one of the largest data transformations in the industry—migrating over 300 petabytes of data to BigQuery—PayPal recognized that an AI-ready data foundation was just the first step of the journey.
“Running AI at PayPal scale means milliseconds matter and no downtime,” explains Manoharan. “Our on-prem infrastructure could not keep up. GPUs were scarce, procurement was slow, and we lacked the elasticity that gen AI demanded.” In addition, fragmented adoption across domains meant teams were building independently in silos, resulting in redundant efforts and draining productivity.
To lead the way for digital commerce, PayPal wanted to create secure, consistent paths to applying AI faster, and more broadly, across the business.
“When you have multiple agent workflows running across a global payments infrastructure, the complexity is enormous. You need one platform that orchestrates everything in one place,” Manoharan says. “The goal was to build a unified, governed AI platform that makes the entire AI life cycle easy and accessible to every engineer.”
When you have multiple agent workflows running across a global payments infrastructure, the complexity is enormous. You need one platform that orchestrates everything in one place. The goal was to build a unified governed AI platform that makes the entire AI life cycle easy and accessible to every engineer.
Srinivasan Manoharan
Director of AI/ML Platform, PayPal
Building a unified backbone for innovation
Partnering with Google Cloud, PayPal is reimagining its technology foundations and infrastructure.
“We built our entire AI stack on top of Google Cloud because it gives us the compute horsepower we need, and the flexibility to scale up as demand arises, and we are now transitioning into the B200 GPUs,” Manoharan says.
With Google Cloud infrastructure, we have achieved five nines of availability for our most critical AI use cases. We serve about 500-plus models in production, handling nearly one billion requests every single day—and the platform doesn’t blink.
Srinivasan Manoharan
Director of AI/ML Platform, PayPal
Central to this approach was Google Kubernetes Engine (GKE), which allowed PayPal to move to a highly elastic architecture, eliminating the risks of traditional, monolithic development and providing the dynamic scaling necessary to power global AI workloads.
“GKE has been transformational,” Manoharan observes. “Before, every release required weeks of regression testing, but after GKE, our release life cycle went from weeks to minutes.” This newfound agility also extends to day-to-day model development and training activities, allowing PayPal’s data scientists and analysts to run 25,000 pipelines every day.
“With Google Cloud infrastructure, we have achieved five nines of availability for our most critical AI use cases,” Manoharan adds. “We serve about 500-plus models in production, handling nearly one billion requests every single day—and the platform doesn’t blink.”
With this high-performance AI infrastructure in place, PayPal’s next priority was to simplify how developers access and use AI models. To help abstract the complexity of managing multiple agents, PayPal built a unified large language model (LLM) stack, which runs open-source models on top of the inferencing stack on GKE and close-sourced models like Gemini on top of Gemini Enterprise Agent Platform. “PayPal needs the flexibility to adopt the best model for each use case every time and swap when something better comes along without rewriting anything,” Manoharan says. “Developers get choice—Claude, Gemini, open source—through one consistent layer and the business gets control. That’s what makes the unified LLM stack so powerful.”
Agent Platform not only provides the digital tools and infrastructure needed to build, scale, govern and optimize agents but developers have access to more than 200 enterprise-ready models through Model Garden. Now, instead of having to reinvent the underlying infrastructure for every new project, developers can grow quickly from idea to innovation while meeting PayPal’s rigorous security and compliance standards.
“Google Cloud handles the infrastructure heavy lifting. For developers, the TCO is zero,” Manoharan says. “Model deployment that used to take weeks now takes minutes. That’s 10,000 engineers across PayPal focused entirely on applying AI for their business problems and not managing infrastructure.”
Shaping the future of agentic commerce
As one of the world’s most trusted payment platforms, processing roughly $1.8 trillion payments annually, PayPal is committed to securing both transactions and financial information. Manoharan underscores that trust remains the most critical hurdle to moving beyond chatbots to production-grade AI.
“We can't put agents into production if we do not have clear visibility into how they behave,” he says. “For agents to work, people have to trust them, and that is our core advantage. PayPal infrastructure ensures every agent-driven transaction is secure, auditable, and protected.”
Using open standards like Model Context Protocol (MCP) and Google’s Agent Payments Protocol (AP2), PayPal is ensuring agent-led transactions remain secure, auditable, and accountable—even if a human isn’t present at the moment of purchase.
MCP allows agents to safely access and communicate with payment systems in a structured, auditable way, logging every action and data exchange. AP2 ensures every action is authorized and verifiable by using cryptographically signed digital contracts, known as Mandates, that specify price limits, timing, and other conditions. Together, Manoharan says these standards are helping PayPal create a trust layer that proactively enforces guardrails and ensures that every decision, action, and interaction is fully visible in real time.
Faster innovation, finance reliability, and an entire workforce empowered with AI—that's what building on Google Cloud has made possible for PayPal.
Srinivasan Manoharan
Director of AI/ML Platform, PayPal
“Today, we support hundreds of internal tools through our MCP layer, enabling agents to take real actions across our internal platforms, not just generate insights and responses,” he states. “Autonomy, guardrails, and transparency operating together—that's what lets us deploy AI in a regulated, customer-facing environment and actually stand behind it.”
The result is that while the payments industry is making strides towards agentic AI, PayPal is ahead of the curve. By deploying AI agents across both customer-facing commerce and internal operations, PayPal is setting a new standard for the future of agentic commerce, automating complex engineering tasks and freeing teams to focus on enhancing experiences for millions of consumers and merchants who rely on the platform every day.
“Faster innovation, finance reliability, and an entire workforce empowered with AI—that's what building on Google Cloud has made possible for PayPal,” Manoharan says.
PayPal has been revolutionizing commerce globally for more than 25 years, creating innovative experiences for consumers and businesses in approximately 200 markets that make moving money, selling, and shopping simple, personalized, and secure.
Industry: Financial services
Location: United States
Products: Google Kubernetes Engine (GKE), Gemini Enterprise Agent Platform, BigQuery, Cloud GPUs
