From PepsiCo to Lowe's: Real Companies Show AI's ROI Is No Longer Theoretical

For years, the hardest question in enterprise AI was simple: does it actually pay off? In March 2026, NVIDIA's annual State of AI report — drawing on more than 3,200 responses across financial services, retail, healthcare, telecommunications, and manufacturing — answered with a clear yes, and backed it with named companies and concrete numbers. The experimentation phase is giving way to measurable returns.

The ROI numbers

The survey found 88% of respondents said AI increased annual revenue in some or all of the business — with nearly a third (30%) reporting gains greater than 10%. On the cost side, 87% said AI helped reduce annual costs, and in retail and CPG, 37% reported cost reductions above 10%. Adoption tracked with scale: 76% of large companies (1,000+ employees) are actively using AI, versus smaller firms still ramping up.

Manufacturing: PepsiCo's digital twins

One of the most concrete examples comes from PepsiCo, which — working with Siemens and NVIDIA — is converting U.S. manufacturing and warehouse facilities into high-fidelity 3D digital twins that simulate entire plant operations. Recreating every machine, conveyor, and operator path at physics-level accuracy lets AI agents test changes virtually and catch up to 90% of potential issues before any physical modification. The early results: a 20% increase in throughput, faster design cycles with nearly 100% design validation, and 10–15% reductions in capital expenditure.

Retail: Lowe's at store scale

Fortune 100 retailer Lowe's built physically accurate digital twins of more than 1,750 stores to speed operations, and used AI to transform 2D product images into precise 3D models in minutes — at a cost of less than $1 per model. It's a vivid example of AI compressing work that was previously slow and expensive into something fast and cheap enough to do at scale.

Healthcare: agents at the bedside

In healthcare, Mona by Clinomic — an onsite assistant for ICU doctors and nurses — consolidates and visualizes patient data in real time. The reported impact is striking: a 68% reduction in documentation errors and a 33% reduction in clinicians' perceived workload. It's a concrete instance of the broader agentic-AI surge the report documents, with telecommunications leading agentic adoption at 48%, followed by retail and CPG at 47%.

What it means

The throughline across these cases is that AI's value is now demonstrable in specific operational metrics — throughput, error rates, capex, cost-per-task — not just in optimistic forecasts. The report also flags the real constraints: data readiness and a shortage of AI experts remain the top barriers to scaling. For any organization, the lesson from the leaders is to target well-defined, measurable use cases, get the data foundation right, and let proven ROI fund the next wave of investment.

Source: NVIDIA, "How AI Is Driving Revenue, Cutting Costs and Boosting Productivity for Every Industry in 2026," March 9, 2026, based on the NVIDIA State of AI surveys (3,200+ respondents).