How they do it, not what they do it with: PepsiCo cracks enterprise AI with novelty

PepsiCo has cracked the code on enterprise AI adoption, but not in the way most companies expect. While businesses worldwide scramble to deploy the latest AI tools, the beverage and snack giant is taking a fundamentally different approach: buying best-in-class technology while fiercely protecting ownership of core business processes.

This strategic balance—embracing external innovation while maintaining internal control—represents a significant departure from the outsourcing-heavy model that dominated enterprise technology for decades. Dr. Athina Kanioura, PepsiCo’s Chief Strategy and Transformation Officer, shared this philosophy at Salesforce Dreamforce 2025, outlining an approach that other enterprises can adapt for their own AI transformations.

Platform partners, process sovereignty

PepsiCo’s AI strategy rests on a simple but powerful principle: the company will partner with technology vendors but never surrender control of its core business operations. “We want to own our core AI-augmented processes, and we will not outsource,” Kanioura explained, emphasizing that PepsiCo’s operating system must remain with company employees rather than external partners.

This isn’t a recent pivot driven by AI hype. PepsiCo has spent years building a comprehensive data-driven foundation that now supports AI applications across five unchanging business priorities: consumer closeness (direct business-to-consumer relationships), commercial excellence (sales and service optimization), operations (logistics and manufacturing), integrated business planning (connecting commercial, financial, and supply chain data), and employee experience (tools, training, and workflow optimization).

The company’s technical foundation reflects this strategic focus. PepsiCo consolidated approximately fifty separate data repositories—collections of information stored in different systems—into one unified global data platform. This consolidation runs on a multi-cloud architecture spanning Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform, with Databricks providing analytics capabilities and a custom application layer called DSX ensuring different systems can communicate effectively.

However, despite embracing cutting-edge technology, Kanioura draws a sharp distinction between what should be purchased from vendors and what must remain under PepsiCo’s direct control. “We buy your products. If we don’t influence your product roadmap, we are not interested,” she stated, describing how this philosophy now shapes relationships with major technology partners including Salesforce, AWS, Microsoft, ServiceNow, Nvidia, and Siemens.

This approach becomes particularly critical as “agentic AI”—artificial intelligence systems capable of taking autonomous actions rather than simply providing recommendations—begins entering enterprise environments. When AI systems can make decisions independently, maintaining control over the underlying business processes becomes essential for managing risk and ensuring outcomes align with company objectives.

Moving beyond the outsourcing model

PepsiCo’s current approach represents a deliberate shift from the heavy outsourcing model that many enterprises adopted over the past two decades. Rather than handing entire functions to external providers, the company now operates on a hybrid model where technology vendors supply building blocks while PepsiCo retains ownership of the blueprint.

This distinction matters at PepsiCo’s scale. The company operates more than 300 manufacturing facilities and thousands of warehouses worldwide. A fragmented technology approach created decision-making bottlenecks and increased operational risk across this complex global network.

To address these challenges, PepsiCo standardized its core enterprise processes to achieve 70 percent commonality across all geographic markets, with the remaining 30 percent reserved for local regulatory requirements and tax considerations. Global process owners now enforce these standards, ensuring consistency while maintaining necessary flexibility.

The company also rejected what Kanioura called the “1000 AI pilots” approach—the common enterprise pattern of launching numerous small-scale AI experiments across different departments. Instead, PepsiCo focused on developing a core set of AI implementations that any business function can reuse, creating economies of scale and reducing complexity.

This centralized approach includes robust governance structures with quarterly oversight involving the general counsel, audit committee, and board of directors, plus a comprehensive Responsible AI policy that establishes guidelines even where regulatory requirements remain unclear.

Industrial-scale workforce development

Technology transformation requires parallel investment in human capabilities, and PepsiCo has approached workforce development with the same systematic thinking it applies to technology platforms. The company launched a Digital Academy providing every employee with foundational knowledge of cloud computing, data analysis, and automation, followed by an AI Academy that opened eighteen months ago.

The learning platform extends beyond traditional corporate training. PepsiCo developed “PepGPT,” a private AI environment with role-based certifications and job-specific curricula. Even truck drivers receive applied AI training focused on dynamic routing, safety protocols, and route optimization tools that integrate with forward and rear-facing cameras and biometric monitoring systems.

Results from a 2023 case study conducted with the Aspen Institute, a nonpartisan policy research organization, demonstrate the program’s impact. The Digital Academy now contains more than 11,000 learning resources and delivered 140,000 completed training modules in its first year, with 600 technical certifications spanning platforms from Azure to DevOps and Power BI.

PepsiCo’s “myeducation” benefit offers employees access to over 100 credentials at no cost, ranging from high school diplomas to university degrees, with the company covering upfront tuition and fees. Enrollment trends toward high-demand digital fields, and the program has delivered measurable business results: participants are nearly twice as likely to receive promotions or role changes, and their attrition rate runs 18 percent lower than the company average.

Lessons for other enterprises

PepsiCo’s approach offers three key insights for other organizations navigating the build-versus-buy decision in AI implementation.

First, enterprises should demand meaningful influence over vendor product development timelines. When a platform powers critical business processes, companies should insist on collaborative design processes and release schedules that align with their operational calendars rather than solely following vendor priorities. PepsiCo makes this collaborative approach a prerequisite for major technology partnerships.

Second, organizations should consolidate experimental AI projects into reusable platforms rather than maintaining numerous isolated pilots. PepsiCo converted approximately 200 individual AI implementations into ten shared services, creating cost efficiencies and simplifying management oversight. This consolidation approach measures success through reusability rather than novelty.

Third, workforce development deserves equal investment priority alongside technology platforms. PepsiCo’s comprehensive learning ecosystem—spanning the Digital Academy, AI Academy, and myeducation program—sits alongside technical infrastructure as a core component of the transformation strategy. The curriculum covers both frontline operational roles and corporate functions, with learning outcomes tied to career mobility rather than simple course completion metrics.

The control imperative

The most significant insight from PepsiCo’s approach transcends specific technology choices or vendor relationships. As Kanioura emphasized, “We want to own the process.” This isn’t a philosophical preference—it’s an operational necessity that enables strategic flexibility.

When enterprises maintain control over their process layer, they preserve the ability to swap AI models, change vendors, and capitalize on technological advances without becoming captive to any single provider’s roadmap or business model. This process ownership becomes particularly valuable as AI capabilities evolve rapidly and competitive dynamics shift in the technology sector.

For enterprises beginning or accelerating their AI transformations, PepsiCo’s model suggests that the most critical decisions involve not which tools to adopt, but rather how to maintain strategic control while leveraging external innovation. The companies that master this balance will be best positioned to adapt as artificial intelligence continues reshaping business operations across industries.