On May 22, Walmart’s US CEO John Furner told investors the company is “becoming AI native.” Not experimenting with AI. Not piloting AI. Native.
That word choice matters. It showed up on a public earnings call, in front of analysts who move billions based on what executives say. Furner did not hedge it. He said Walmart is using AI to serve customer needs that previous technologies could not meet. Then he backed it with numbers.
The numbers are not subtle
Sparky, Walmart’s AI shopping agent, doubled its weekly active users in Q1. Not over a year. In a single quarter.
Average order value for Sparky users is 35 percent higher than for non-Sparky users. Units purchased through Sparky grew more than 4x sequentially. Response quality improved 40 percent this year alone.
These are not engagement metrics from a chatbot experiment. They are commerce metrics. Revenue per user. Units moved. Operational impact measured the way a retailer measures everything: did more product leave the shelf?
The answer is yes, at significant scale.
What Walmart actually did
Sparky is now live across the app, web, and in-store experiences. It handles personalized replenishment, meal planning, and product recommendations based on real-time inventory, local pricing, and delivery speed.
This is not a search bar that answers questions. It is a system that knows what you bought last Tuesday, understands that you are feeding four people, checks whether chicken thighs are in stock at your nearest store, and suggests a recipe based on what is on sale this week. Then it adds everything to your cart in one action.
The capability itself is impressive. But the operational decision behind it is what matters. Walmart did not build a better chatbot. They redesigned the shopping workflow around an agent that holds context across sessions, integrates with inventory and logistics systems, and makes decisions on behalf of the customer.
That is an operations problem solved through AI. Not a technology showcase.
Why this is a gap story
Walmart serves 255 million customers per week across 10,500 stores in 19 countries. When a company that size says it is AI native and proves it with a 100 percent growth curve in one quarter, the competitive signal is not “they have good AI.” The signal is “they restructured how a quarter-billion people buy things, and they did it while everyone else was still running pilots.”
Consider what happens to a mid-size retailer watching this. Walmart’s AI agent is not just faster than a search bar. It is stickier. A customer whose meal plans, purchase history, and preferences live inside Sparky has a switching cost. They are not going to open a competitor’s app and start from scratch. That is a moat built on operations, not on model quality.
Six months from now, Sparky will have another two quarters of user data compounding its recommendations. The gap between what Walmart offers and what a non-AI-native retailer offers will be measurably wider. Not because Walmart has a better model. Because they built the workflow that turns a model into revenue.
The question for your business
You do not need to be Walmart. You do not need 2.1 million employees or a $600 billion market cap. But the pattern here applies regardless of size.
Walmart did not wait for perfect AI. Sparky’s response quality improved 40 percent this year, which means it was 40 percent worse when they launched it. They shipped it early, integrated it into real workflows, measured what happened, and improved. The AI got better because it was in production, not because they waited for it to be ready.
The companies pulling ahead right now share this trait. They are not running the best models. They are running real workflows that include AI as a participant, not as a sidebar experiment. They are measuring output in the same units they have always used: revenue, units, customer retention, time saved.
If your organization is still debating which AI tool to buy, or waiting for someone to prove the ROI before you start, Walmart just told Wall Street what the ROI looks like. One hundred percent user growth. Thirty-five percent higher order value. Four times the units.
Those are not demo numbers. Those are Q1 results from the world’s largest retailer.
The question is not whether AI works in your industry. The question is how much distance the companies already running it will put between themselves and everyone else by the time you start.