Guide
Ocado's robot grocery warehouses: what Kroger, Sobeys and Coles filings confirm
Ocado's customer fulfilment centres (CFCs) are large automated warehouses where robots move on a storage grid to bring grocery totes to pickers, and increasingly to robot arms, for home delivery orders. The retailers that pay for them now tell two different stories in their own filings. Kroger closed three Ocado-powered CFCs. Ocado's reports name five remaining Kroger CFCs and the three closed ones (eight that operated, a Servospace count). Kroger also cancelled a planned site in Charlotte, N.C., and booked $2.5 billion of impairment and related charges. Its annual report shows that the Ocado robotic equipment leases on its balance sheet, $926 million at February 1, 2025, were carried at zero by January 31, 2026. That zero is a company-wide Ocado lease figure; Kroger does not publish it by site. Servospace reads it as covering the five sites Kroger kept. Sobeys' parent Empire closed its Calgary CFC less than three years after opening, kept a Vancouver site paused and booked a C$746 million charge. In Australia, Coles reported that its two CFCs delivered positive EBITDA in their second year of operation, with sales growth above 30%.
Both retreating retailers used the same word for where the sites still make sense: density. Kroger says that "in geographies where Kroger sees higher density of demand and better cost structure, the Company continues to evaluate performance and sustainability of automated fulfillment." Empire keeps Toronto and Montreal as "areas with high density demand." For a grocery or distribution center lead, the practical reading is that the customer record does not settle whether the robots work. It settles that a CFC's economics depend on enough orders inside its delivery area, and that when those orders are missing the retailer books the larger write-down. Kroger and Empire took charges of $2.5 billion and C$746 million. Kroger says its charges included a cash payment to Ocado of about $350 million, and Empire's charge includes unnamed one-time cash contract termination payments; Ocado says it received a £17.7 million termination fee for the Calgary closure. Ocado also recorded £125 million of depreciation and impairment on its own books related to Kroger's network optimisation and the Calgary closure, according to its half-year report.
Analysis based on Kroger's Form 8-K and 8-K/A (November and December 2025), Form 10-Q for the quarter ended November 8, 2025, Form 10-K for the year ended January 31, 2026 and Forms 10-Q for the quarters ended May 23 and August 15, 2026; Empire Company's January 28, 2026 release and its MD&A for the quarter ended January 31, 2026; Coles Group's 2026 full-year results presentation (August 25, 2026) and 2026 annual report; Ocado Group's 2025 annual results (February 26, 2026), 2026 half-year results (July 16, 2026), partner pages and Coles news post; and reporting by The Globe and Mail, all retrieved November 17, 2026. AI assistance was used in research and drafting, and the piece was checked by a separate AI fact-check and AI editorial review. Servospace did not interview the companies or visit any site. Kroger figures are U.S. dollars, Empire figures are Canadian dollars and Coles figures are Australian dollars. Calculations are Servospace arithmetic on published figures and are labeled as such. Last reviewed November 17, 2026.
In this analysis
- The short answer
- What kind of evidence each number is
- Kroger: three closures and robot leases at zero
- Sobeys: one of three closed, a fourth paused
- Coles: two sites, positive EBITDA in year two
- What Ocado's own counts show
- Where the two accounts diverge
- What a grocery buyer should ask
- Limitations and open questions
- What would change this analysis
The short answer
- Kroger (United States). Ocado's reports name five remaining Kroger CFCs and three closed ones. Servospace counts eight that operated. The three closed sites, in Pleasant Prairie, Wis., Frederick, Md., and Groveland, Fla., closed in January 2026 because they "had not met operational or financial expectations," Kroger's 10-K says. A planned Charlotte site was cancelled. Kroger paid Ocado about $350 million. Five sites remain in Monroe, Ohio, Dallas, Atlanta, Denver and Detroit, with Phoenix still planned. Ocado's 2018 commitment with Kroger was for capacity equivalent to 20 CFCs, according to Ocado's partner page.
- Sobeys (Canada). Empire closed its Calgary CFC and an Edmonton support facility in January 2026, kept development of a Vancouver CFC paused, and recognized a C$746 million impairment and related costs. Two CFCs remain, in the Greater Toronto and Montreal areas.
- Coles (Australia). Coles' results presentation says its two CFCs, in Melbourne and Sydney, were EBITDA positive in their second year of operations, with CFC sales growth of more than 30%, ahead of total supermarkets eCommerce growth of 26.4%. Coles does not publish the EBITDA amount.
- Ocado's network. At May 31, 2026, Ocado counted 23 CFCs and four smaller Zoom sites live worldwide with 113 live modules, down from 25 CFCs and 119 modules a year earlier. Eleven of the 113 are capacity at Ocado Retail sites where operations have ceased but fees are still charged.
What kind of evidence each number is
The same event reads differently depending on who reports it. Kroger and Empire are the customers paying for the systems and file with securities regulators. Ocado is the vendor, and its counts measure the capacity it bills for. The table keeps the labels attached.
| Source | Who is speaking | Establishes | Does not establish |
|---|---|---|---|
| Kroger 8-K, 10-Q and 10-K | Customer, in securities filings | Closures, the charge, the Ocado payment, the lease write-down, Kroger's stated reasons | Sales, volume or profit for any single CFC |
| Empire MD&A and release | Customer, in securities filings | The Calgary closure, the C$746 million charge, carrying and recoverable amounts, expected savings | Sales or profit for Toronto or Montreal |
| Coles results presentation | Customer, in an exchange filing | Positive CFC EBITDA in year two, CFC sales growth above 30%, CFC NPS ahead of total online | The EBITDA amount, order volume, utilization or capital cost |
| Ocado results and partner pages | Vendor | Site and module counts, fees received, closure fees, delivery dates it guides to | How profitable a site is for the retailer |
| The Globe and Mail | Press | Opening timeline and context for Calgary and Vancouver | Anything not attributed to a company |
Kroger: three closures and robot leases at zero
Kroger announced the change in a Form 8-K on November 18, 2025. It said it would close certain fulfillment centers and expected impairment and related charges of about $2.6 billion "as a result of these closures and the rest of the automated fulfillment network not meeting financial expectations." The charge covered the whole network, not only the closed sites.
An 8-K/A on December 5, 2025 added that the charges included a cash payment to Ocado of about $350 million. Kroger's 10-Q for the quarter ended November 8, 2025 put the charge at $2,585 million and named the three closures and the Charlotte cancellation. Its 10-K for the year ended January 31, 2026 gives the final figure: $2,497 million, or $1,908 million after tax.
Servospace reading: each tested facility failed recoverability. The 10-K explains how the charge was set. Kroger treated each automated fulfillment facility as a separate asset group, and "the recoverability testing indicated the carrying value of each asset group exceeded the sum of undiscounted cash flows expected to result from the use and eventual disposition of the assets." The asset groups were then written down to estimated fair value. Read literally, "each asset group" includes the five CFCs Kroger kept open. That is Servospace's reading of the filing's wording; Kroger does not list the charge by site.
The Ocado equipment leases went to zero. Kroger accounts for its Ocado agreement as a lease of "the robotic equipment used to fulfill customer orders." The 10-K says Kroger had $926 million of net finance lease assets related to the Ocado agreement at February 1, 2025, and that "due to this impairment, the Company did not have any net finance lease assets as of January 31, 2026." The impairment of finance lease assets in 2025 included $948 million related to the automated fulfillment network. Kroger still showed finance lease liabilities of $342 million current and $243 million non-current at year end in the same note, down from $104 million and $781 million, "primarily due to" the Ocado payment.
A zero carrying value is an accounting measure. It does not mean the five remaining sites stopped running, and Ocado says they continue to operate. It does mean Kroger carried those Ocado finance lease assets at nil after the impairment. The 10-K says recoverability failed because carrying value exceeded undiscounted cash flows from use and disposition, then asset groups were written down to estimated fair value. Fair value for lease assets used a discounted cash flow of estimated market rent, not a published per-site profit figure. Kroger does not say the remaining buildings or owned equipment are worth nothing, and it still reported Ocado-related finance lease liabilities.
How the $350 million moved. Kroger recorded $246 million of the payment as a repayment of finance lease obligations and $104 million in operating cash flow. Ocado's 2026 half-year results say it recognized £260.1 million of revenue for the termination fee, plus £67 million of accelerated advance receipts. Earlier, in the second quarter of 2025, Ocado drew the entire remaining $152 million from a letter of credit under the partnership agreement, according to Kroger's 10-K.
What Kroger says now. The 10-K says Kroger "continues to evaluate performance and sustainability of automated fulfillment" in "geographies where Kroger sees higher density of demand and better cost structure," and that the closures "are expected to have a positive effect on eCommerce operating profit and a neutral effect on identical sales without fuel." It also lists the CFCs as a risk: "We have closed certain customer fulfillment centers because they have not been meeting operational and financial expectations."
Kroger does not report CFC sales separately. Its 10-Q for the quarter ended August 15, 2026 says eCommerce sales rose 14% in the quarter, "led by increased demand for Delivery solutions," and 20% excluding "fulfillment center exits in markets where Kroger does not operate stores," the sale of Vitacost.com and the end of Ship Marketplace. Kroger does not say which sites the exits were. The same filing says "our eCommerce business, including third-party media revenue, was profitable in the second quarter and first two quarters of 2026." That covers all of Kroger's eCommerce plus media revenue, not the remaining CFCs. Depreciation fell as a share of sales, which the filing attributes "primarily" to the fulfillment network closures. Servospace found no paragraph on the Ocado agreement in that 10-Q.
Phoenix keeps moving. Kroger's remaining planned CFC is in Phoenix, the first to use Ocado's freezer automation. Ocado's 2025 annual results in February 2026 guided to a Phoenix go-live in fiscal 2026 or 2027. Its July 2026 half-year results moved Phoenix to fiscal 2027 or 2028. Ocado's fiscal year ends around November 30.
A resale. Ocado said in July 2026 that it had "successfully agreed the sale of its technology in one of the recently closed Kroger CFCs to a large U.S-based logistics company, with ongoing support for the maintenance of that equipment." It did not name the buyer or the site.
Sobeys: one of three closed, a fourth paused
Empire announced on January 28, 2026 the "immediate wind-down and closure" of its Alberta e-commerce facilities, a CFC in the Calgary area and a smaller support facility in Edmonton. It said the Alberta operations "have not met the company's financial expectations, largely due to the Alberta grocery e-commerce market's size and expansion being smaller than originally anticipated." The Vancouver-area CFC stays paused, "as previously communicated." The Globe and Mail reported that the Calgary centre closed less than three years after it opened and that Empire paused Vancouver in 2024.
Empire's MD&A for the quarter ended January 31, 2026 gives the accounting detail:
| Empire item, quarter ended January 31, 2026 | C$ |
|---|---|
| E-commerce impairment and related costs | 746 million |
| Carrying amount of the cash-generating units before impairment | 724 million |
| Estimated recoverable amount | 126 million |
| Impairment of right-of-use assets, including 55 million of decommissioning costs | 409 million |
| Impairment of property and equipment | 206 million |
| Impairment of intangibles | 39 million |
| Expected annualized e-commerce benefit | about 95 million |
Servospace arithmetic: the recoverable amount was about 17% of the carrying amount, so roughly 83% of the tested value was written off. The listed impairment lines sum to C$654 million. That sum already includes C$55 million of decommissioning inside the C$409 million right-of-use line. The remaining gap to C$746 million is the MD&A's other direct closure costs, such as severance and "one-time cash contract termination payments." Ocado's half-year report says it received a £17.7 million termination fee for the Calgary closure.
Empire says the C$95 million benefit comes "largely" from "a reduction of operational losses from its Calgary CFC and reduction in future depreciation," and that about one-third will be reinvested. It keeps Toronto and Montreal to support its e-commerce strategy in "these areas with high density demand." Empire had already ended its mutual exclusivity with Ocado early, taking a C$12 million non-cash charge in the first quarter of fiscal 2025. Its combined e-commerce sales from Voilà (including curbside pickup), IGA.net, ThriftyFoods.com and partnerships with Instacart and Uber Eats rose 10.3% in the quarter, "primarily driven by growth in third-party partnership sales and continued sales growth for Voilà."
Ocado's Sobeys page says Ocado is adding its Swift Router software for same-day and short-lead-time orders to the two remaining sites, and that Sobeys uses Ocado's in-store fulfilment software in 87 stores.
Coles: two sites, positive EBITDA in year two
Coles is the clearest customer-reported success. Its 2026 full-year results presentation, for the year to June 2026, says the CFCs were "EBITDA positive in their second year of operations." It reports CFC sales growth of more than 30%, ahead of total supermarkets eCommerce growth of 26.4%, and a CFC Net Promoter Score 710 basis points ahead of total online. A caption in the presentation dates the opening of the Truganina, Victoria CFC to September 2024. Ocado's news post on the results says Coles opened both CFCs, in Melbourne and Greater Sydney, in 2024.
Coles credits the result to volume and to added automation. Its 2026 annual report says "strong volume growth was supported by expanded catchment areas and the introduction of same day delivery from the CFCs," that it installed on-grid robotic pick arms and auto frame loading, and that store-fulfilled Click & Collect gained capacity "as same-day orders in Melbourne and Sydney transitioned to the CFCs." Coles' presentation also lists auto bagging; Ocado's post repeats it.
Two limits matter. Coles does not publish the EBITDA amount, so "positive" could mean a small margin. And EBITDA excludes depreciation and lease costs, which are exactly the lines Kroger and Empire wrote down. A positive EBITDA shows the sites cover their operating costs; it does not show they repay the capital.
The difference in setting is visible in the filings themselves. Coles placed two sites in Australia's two largest cities and moved same-day orders there from its stores. Kroger's 2026 filings refer to "fulfillment center exits in markets where Kroger does not operate stores," and Empire's closure was in a province whose online grocery market it calls smaller than expected.
What Ocado's own counts show
Ocado reports capacity in modules, a unit of grid capacity it bills for. Its counts are vendor figures, but they are dated and consistent, which makes them a useful timeline.
| As of | Live sites | Live modules | Notes (Ocado's wording) |
|---|---|---|---|
| May 31, 2025 (1H25) | 29: 25 CFCs and four Zooms | 119 | Comparative in 2026 half-year results |
| Nov 30, 2025 (FY25) | 30 group live sites (26 CFCs, four Zooms). Technology Solutions: 27 where partners were actively trading (25 CFCs, two Zooms) | 122 group; 108 actively trading | Do not compare the 27 actively trading sites with the 29 and 27 group live-site rows; use the group total of 30. 122 includes 14 modules where Ocado Retail has ceased operations but is still charged; Morrisons deliveries ended at Erith (5 modules) |
| May 31, 2026 (1H26) | 27: 23 CFCs and four Zooms | 113 | Includes 11 modules at ceased Ocado Retail sites; Technology Solutions had 24 sites and 102 modules in operation |
Ordered capacity is shrinking too. At November 30, 2025, Ocado had 163 modules ordered at sites in operation and 35 at sites under construction for AEON, Lotte and Kroger. At May 31, 2026, those figures were 144 and 26. Ocado says it expects about 10 new modules in fiscal 2026 and about 10 in fiscal 2027, "offsetting 12 closed in early 2026."
Ocado says the four closed Kroger and Sobeys CFCs generated a contribution of £22 million in the first half of 2025 and £8 million in the opening weeks of the first half of 2026. Separately, it says like-for-like recurring fees, after removing business-as-usual fees from those four sites of £27 million in 1H25 and £9 million in 1H26, grew 5% to £222 million; including those sites, recurring fees fell 3% to £231 million.
Ocado also reports progress at newer partners: international CFC volumes up 27% year on year, On-Grid Robotic Pick in 14 CFCs for eight partners, and "particular progress among recent partners who have gone live, in Spain, Poland, Japan and Australia." Those are vendor statements. Of those markets, only Coles' figures above come from the retailer itself in the sources reviewed here.
Where the two accounts diverge
Ocado's Kroger page says its teams are supporting "significant progress in operational efficiency, volume growth, and increasing same day availability, resulting in a significant improvement in financial performance" across the remaining Kroger CFCs; its February 2026 annual results use nearly the same words about "the underlying financial performance of the CFCs." Kroger's 10-K, filed March 31, 2026, says the company "continues to evaluate performance and sustainability" of automated fulfillment, and its impairment test found each facility's carrying value above its expected undiscounted cash flows.
These are not necessarily in conflict. Ocado may be describing improvement from a low base, and an impairment resets the cost base so that the same sites can look better afterward, because depreciation on written-down assets falls. Kroger's 10-Q says exactly that depreciation fell after the closures. But a reader should not take Ocado's description of "significant improvement" as the customer's verdict. The customer's verdict, so far, is a company-wide write-down to zero on the leased Ocado equipment and an open evaluation.
What a grocery buyer should ask
| Question | Why it matters | Where the public record stands |
|---|---|---|
| How many orders per week will the catchment produce, and by what year? | Both retreating retailers named demand density as the deciding factor | Kroger and Empire name density; neither publishes the threshold |
| What utilization of design capacity does each site run at? | Fixed capacity fees and depreciation are spread over volume | Ocado's half-year report puts Ocado Retail's network at 103% of original design capacity; no partner in this analysis reports a figure |
| What does it cost to exit a site, and when? | Kroger paid about $350 million for three closures and one cancellation; Ocado recorded £17.7 million from Sobeys | Contract terms are not public |
| Can the equipment be redeployed or resold? | A closed site's grid is otherwise sunk cost | Ocado agreed the sale of one closed Kroger site's technology to an unnamed logistics company |
| What happens to exclusivity? | Exclusivity limits other options in the same market | Empire ended its exclusivity early for a C$12 million charge; Ocado says exclusivity has ended in most markets |
| Does the site serve existing store areas? | Store-adjacent demand and same-day orders moved from stores helped Coles | Kroger's 2026 filings exclude "fulfillment center exits in markets where Kroger does not operate stores" from an eCommerce growth figure. Kroger does not name those sites. |
Limitations and open questions
- No per-site results. None of the three retailers publishes sales, orders, labor hours or profit for an individual CFC. The comparison here is between disclosed decisions and disclosed aggregates.
- EBITDA is not payback. Coles' positive EBITDA excludes the capital and lease costs that drove the North American write-downs.
- Reading "each asset group." The statement that each Kroger facility failed the recoverability test is Servospace's reading of the 10-K's wording. Kroger has not broken the charge out by site.
- Vendor counts. Module and site counts are Ocado's own, defined by what it bills; they include closed Ocado Retail capacity that is still charged.
- Partners not covered. This analysis does not cover AEON, Lotte, Bon Preu, Auchan Poland, ICA, Groupe Casino, Asda (announced in May 2026 as a new UK partner, due to go live in fiscal 2027) or Ocado Retail with partner-issued data. Ocado Retail is 50% owned by Ocado and is not an arm's-length customer. For how Ocado measures its own picking, see the robot piece-picking analysis.
- Grid storage is not one market. Ocado's grid is a grocery home-delivery system; comparisons with general warehouse grids belong with the AutoStore installed-base analysis and the Exotec Skypod analysis. For another case where a retailer's filings and a vendor's backlog tell different stories, see the Symbotic and Walmart analysis.
What would change this analysis
The next checkpoints are Kroger's results for the quarter ending in early November 2026 and its fiscal 2026 10-K, Ocado's fiscal 2026 annual results, Empire's fiscal 2026 annual report, Coles' 2027 half-year results, and any go-live of AEON's second Tokyo CFC, Lotte's Busan CFC or Kroger's Phoenix CFC. Any customer-published per-site volume, utilization or profit figure, a further Kroger impairment, or a new U.S. partner announcement would trigger a dated update. This guide will be reviewed when one of those is published, or within 90 days. For related coverage, see warehouse robot news.
How this analysis was done
Customer filings are treated as the customer's own statements. Ocado documents are treated as vendor statements, including where they describe a partner. Press reports are cited only for what they attribute to a named company. Quotes are taken from the primary documents linked above. Calculations are simple divisions or sums of published figures and are labeled as Servospace arithmetic. Corrections can be sent through Servospace's contact page.