Micro-fulfillment is one of my favorite topics to write about. Because of the amount of research I’ve conducted on the topic, and my hands-on experience with robotics and automation, I’m frequently referred to as an “expert in micro-fulfillment and robotics.” The purpose of this article is to educate and inform.
Every year at GroceryShop, the automation vendors on the floor tell a version of the same story: robots, speed, lower cost to serve. Most of the audience nods along without asking the question that matters, which isn’t whether a system works, it’s who built it and why. That question is about to be tested at the highest-profile stage in the industry, because Walmart, the largest grocery retailer in the country, has placed its bet on a vendor that didn’t design the system it’s now piloting.
I want to walk through that decision in detail, because it’s the clearest live case study the grocery industry has right now for a question every retailer attending GroceryShop will eventually face: when you automate the store, do you buy proven, purpose-built grocery technology, or do you buy an industrial automation company’s best attempt at retrofitting something it acquired secondhand?
Walmart and Symbotic’s Unique Relationship
Walmart’s automation story starts with Alert Innovation, an independent robotics company that designed and built the original micro-fulfillment system now at the center of Walmart’s plan. Walmart could have partnered with AutoStore early on instead, but several inexperienced team members responsible for Walmart’s store technology convinced themselves that AutoStore couldn’t meet Walmart’s needs. Nonsense. I spoke with the individuals who made that call, and it was clear to me they were incapable of understanding what they were evaluating.
H-E-B, considered to be the best grocery retailer in the United States, ran the rigorous evaluation Walmart didn’t. Every available MFC was evaluated, including TakeOff Technologies, Fabric, Alert Innovation, Dematic, and AutoStore. AutoStore was chosen. Full disclosure: I assisted with the selection process. As a former advisor to TakeOff and Fabric, and an expert on the topic of Alert Innovation and AutoStore, the only logical choice was AutoStore. H-E-B has installed a dozen AutoStore systems across its network. Walmart should have partnered with AutoStore, and the fact it didn’t is embarrassing.
What most coverage of Walmart’s actual choice, Alert Innovation, has missed is where that technology traces back to. Its founder, John Lert, has been trying to automate the grocery store since 1994. In 2007, Lert founded CasePick Systems with funding from Rick Cohen, owner of C&S Wholesale Grocers, to build a mobile robotic shuttle system for warehouse automation. Cohen bought out full ownership of the company in 2009 and renamed it Symbotic. Rick Cohen is Symbotic’s founder, chairman, and CEO today, not Lert. Lert left in 2011, still convinced the bigger opportunity was a true three-dimensional shuttle rather than the system Cohen had taken over, and founded Alert Innovation in 2013 to build it.
Lert frequently contacted me and asked for my opinion. During a discussion about AutoStore and his Alphabot system, I asked him a pointed question directly: is the Alphabot better than AutoStore’s? Lert replied, “No it’s not. AutoStore is a better system.” I have tremendous respect for John Lert. His honesty was refreshing, and we were able to discuss opportunities for how his platform could eventually become better than AutoStore.
Things rapidly changed for Alert Innovation in 2019, when Walmart contracted the company to install an Alphabot in a store in Salem, New Hampshire. Additional Alphabots followed in other Walmart stores, and in 2022, Walmart acquired Alert Innovation outright. In January 2025, Walmart sold that unit to Symbotic for $200 million, alongside a $520 million commercial agreement under which Walmart is now paying Symbotic to turn the system into a working store-level fulfillment product, branded SymMicro.
Symbotic’s own core expertise, built through its work with Albertsons, C&S Wholesale Grocers, and Target, is large-scale industrial distribution center automation, palletized freight moving through massive facilities. Symbotic is truly a leader in warehouse automation. SymMicro is the company’s first attempt to take that warehouse expertise and shrink it down into a live grocery store, using a system it didn’t design and only recently acquired secondhand.
I’ve used this comparison before and it holds up: it’s the equivalent of Ford deciding to build a new muscle car by taking an F-150 pickup truck and customizing the drivetrain, rather than purpose-designing a car from the frame up. The F-150 is an excellent truck. It was never meant to be a Mustang.
Rick Cohen confirmed on Symbotic’s own August 2026 earnings call that the first SymMicro installation, the single pilot store running today, wouldn’t be fully online until roughly six months out, into early 2027. He described Walmart’s own evaluation process as staged and deliberate: first an “overbuilt prototype,” then a “redesigned version with lower cost, smaller size and refined functionality,” before Walmart will even consider committing to scale. The actual 400-store order isn’t expected until early 2028, by Cohen’s own account on that call.
That’s not caution for its own sake. That’s a company still validating, three years after acquiring the underlying system, whether a retrofitted platform can work at all. Now compare that timeline to the alternative Walmart passed on.
Ocado Is the Leader in Grocery Automation
Ocado took the opposite path from day one. It spent 25 years as an online grocery operator first, building automation to solve its own operating problem, low margins, high SKU variability, multiple temperature zones, before it became a technology vendor to other grocers.
The scale is worth putting in front of this audience directly, because it’s easy to underestimate from outside the category. In the first half of 2026 alone, Ocado’s technology delivered more than 350 million picks across 14 Customer Fulfilment Centers and eight partners worldwide, with international volume growth of 27% year over year.
Individual sites run fleets of over 1,000 robots, process millions of items a week at accuracy rates above 99.9%, and operate pick stations exceeding 600 items per hour, nearly an order of magnitude faster than manual grocery picking. The newest hardware, the 600 Series grid robot, is five times lighter than its predecessor and more than half 3D-printed, built specifically to run on lighter, faster-to-install grids inside existing buildings rather than requiring a purpose-built facility.
Ocado’s newest product, Store Based Automation, SBA, is a compact version of that same proven technology, sized to fit inside or beside an existing store instead of requiring a standalone 350,000-square-foot Customer Fulfilment Center. It’s not a new company’s first attempt at grocery automation. It’s a 25-year grocery automation leader’s answer to the exact problem this room is trying to solve: how to automate an existing footprint without a nine-figure standalone build.
None of that means Ocado has been beyond reproach, and I want to be upfront about that. I’ve been critical of Ocado in several articles I’ve written. I publicly called for Ocado to build a micro-fulfillment system in 2020, 2022, and early 2025, and I made the argument that it was inexcusable for Ocado to remain focused on building Customer Fulfillment Centers while the market moved toward smaller, store-based formats. Ocado formally announced its intention to build Store Based Automation, its micro-fulfillment system, on November 19, 2025.
What About Kroger?
Anyone who’s followed Ocado’s U.S. story is already thinking about Kroger, and it deserves a direct answer rather than a dodge. Kroger closed three of its Ocado-powered CFCs in early 2026, a closure for which Ocado paid Kroger more than $250 million in compensation. That’s real, and it happened.
What it wasn’t was a technology failure. It was a site-selection and demand-forecasting failure: those facilities were built in locations without the order density to make a 350,000-square-foot automated CFC economical, a point I can speak to directly. I’m the person who recommended Kroger acquire or partner with Ocado in the first place, and I designed a different strategy for how Kroger should have deployed Ocado’s technology than the one Kroger actually implemented. Had Kroger followed my recommendations, far fewer CFCs would have been built.
The same underlying Ocado system that struggled at Kroger is delivering the opposite result for Coles in Australia, installed at sites matched to real demand. Coles reported positive EBITDA in only its second year of operation, according to its own August 2026 results, driven by rising volumes and on-grid robotic picking.
Same technology, different outcomes depending on the site, and that’s the real lesson, not a verdict that SBA should replace CFCs everywhere. A large-format CFC still makes sense where order density supports it, which is exactly why Coles kept investing in its two Australian CFCs and why Kroger itself isn’t abandoning CFCs, its November 2025 restructuring explicitly kept them in its highest-density markets. The failure wasn’t the CFC model in general. It was building standalone, capital-intensive facilities in locations that couldn’t generate the order volume to justify them. Kroger’s own hybrid approach, keeping CFCs where density supports them while piloting capital-light, store-based automation in high-volume geographies that don’t, is the more accurate model for the attendees at GroceryShop to take away: match the fulfillment format to the market, don’t pick one automation architecture and force every site into it.
Lotte Shopping in South Korea is the case worth watching as this plays out globally. Ocado signed its partnership with Lotte in November 2022, broke ground on the first Customer Fulfilment Centre in Busan in December 2023, and brought it live in August 2026, an Ocado-powered CFC now serving roughly 4 million households across Busan and the wider Yeongnam region. The site introduced several Ocado innovations for the first time anywhere in the world, including a fully automated freezer operating down to -25°C, on-grid robotic picking, and support for South Korea’s “dawn delivery” standard, fresh orders on the doorstep before 7am. Lotte and Ocado plan to build as many as six CFCs across the country by 2030, targeting 5 trillion won in online grocery sales, with a second site already planned for Goyang to cover the Seoul metro area. As Ocado’s own fulfilment strategy has evolved to combine large CFCs with compact Store Based Automation depending on market density, Lotte’s remaining rollout is a natural candidate to follow that same hybrid model, deploying full CFCs in its highest-density regions and SBA in the markets that don’t justify a standalone facility, though Lotte has not yet publicly confirmed SBA as part of its specific buildout plan.
The Question Every Retailer Should Be Asking
This isn’t really a Walmart story. It’s a category story, and it’s why I’m putting it in front of this specific audience. Every retailer walking the show floor at GroceryShop is going to face some version of the same decision in the next 24 months: automate the store and choose which vendor’s technology to bet on.
The instinct in the grocery industry is often to default to the vendor with the existing relationship, the familiar name, the lower perceived switching cost. Walmart’s SymMicro decision is a live example of exactly that instinct, choosing continuity with a known vendor over a harder, more rigorous evaluation of who has the deepest grocery-specific expertise.
I recommend every retailer evaluating store-level automation run the comparison Walmart hasn’t yet run publicly: a real, side-by-side pilot between a purpose-built grocery automation platform and a retrofitted industrial system, under identical conditions, before committing capital at scale. Ocado and AutoStore should both be on that list. I’ve said publicly that I believe Walmart CEO John Furner will choose Ocado’s SBA over SymMicro once that comparison happens. Whether or not that prediction holds for Walmart specifically, the underlying discipline, pilot before you commit, and weigh the vendor’s track record as heavily as its pitch, is the right approach for every company assessing automation.
Twenty-five years of grocery-specific engineering is not a marketing claim. It’s a track record you can pilot against. Make vendors prove it before you sign the contract that locks in your automation strategy for the next decade.
Based on multiple discussions with sources across the grocery industry, I believe that Walmart, H-E-B, Ahold Delhaize, Kroger, Amazon, Albertsons, and Target are all currently in discussions with Ocado about its SBA solution, or soon will be. I’m not claiming any of these companies will sign with Ocado, only that the conversations are happening.