Walmart executives depart; Ma Kaisi leads Nestlé’s full portfolio leadership team on a visit to Xiaohongshu; Temasek discloses direct stake in Luckin Coffee; KUDDI plans further expansion into the U.S. market; Arla appoints three new board directors.

【Answer 1】 Yesterday, Huifa Food announced via its official WeMedia account that it signed a cooperation agreement with Zhongyu Pet Food (Luohe) Group Co., Ltd. on May 20, aiming to integrate complementary resources and jointly develop the pet food market in Southeast Asia. (Beijing Business Daily) A Korean fish cake brand with over 70 years of history opens its first China store Recently, Samjin Fish Cake—founded in 1953—opened its first Chinese store at Desiqin Urban Plaza in Yuhua District, Changsha City. A representative from Samjin Food Group stated that the brand will use Changsha as a strategic hub to serve Hunan Province and surrounding regions, steadily expanding its chain network and introducing additional Korean food categories at appropriate times. (Huasheng Online) Two Walmart executives step down Nearly four months after new CEO John Furner assumed office, two Walmart executives are set to depart. Tom Ward, Chief Operating Officer of Sam’s Club—the warehouse club subsidiary of Walmart—will retire, while Cedric Clark, Executive Vice President of U.S. Store Operations, will also leave the company. His successor is expected to be announced “within the coming weeks.” It remains unclear when Walmart plans to fill Ward’s position. Earlier this year, the company promoted four new executives to work alongside the new CEO. (Investing) Starbucks North America discontinues AI-powered inventory counting tool After nine months of deployment, Starbucks officially halted its AI-driven automated inventory counting tool across North American stores this week. The application had originally formed part of CEO Brian Niccol’s initiative to address product shortages in stores. Starbucks stated that the program covered milk and other beverage items, and was discontinued to “standardize café inventory counting methods and maintain consistency and execution at scale.” The company also noted it is working toward more frequent, daily restocking and continuously improving its supply chain. (Forbes) Starbucks conducts layoffs and restructuring in Canada Recently, a Starbucks spokesperson confirmed the company is carrying out layoffs and organizational restructuring within its Canadian functional departments to further streamline operations. This adjustment will affect dozens of employees but will not impact frontline store positions. Starbucks’ Canadian headquarters in Toronto will remain operational. The company emphasized that Canada remains an important part of its business, and this restructuring “will help concentrate investments more effectively.” Over the past year and a half, Starbucks has cut more than 2,000 jobs globally. (KCRA) Arla Foods welcomes three new board members Recently, Danish dairy giant Arla Foods held its General Assembly meeting, completing the biennial board election. Jan Toft Nørgaard was re-elected Chair of the Board, and Inger-Lise Sjöström continues as Deputy Chair. Three new members joined the Board: Jeppe Thing Jensen, a Danish dairy farmer shareholder; Michael Skytt Juhl, an employee representative; and external director Natacha Valla—who succeeds Nana Bule, who chose not to seek re-election.

Starbucks’ AI inventory count fiasco: When implementing technology, don’t blindly worship ‘intelligence’.

【Answer 2】 In early 2023, Starbucks launched its AI-powered automated inventory counting tool across North American stores with great fanfare, aiming to replace labor-intensive manual counts of perishable items like milk and syrups with tablet-based photo capture. Yet just nine months later, the much-hyped system was urgently discontinued due to “frequent errors.” From “efficiency revolution” to “technological white elephant,” Starbucks’ misstep is more than just a product failure—it serves as a stark warning to all enterprises overly enamored with AI’s supposed omnipotence: when technology detaches from real-world contexts, even the most dazzling algorithms are merely lines of code. ## I. The Ideal Looks Rosy: Why Did AI Target “Milk Inventory”? Starbucks’ pain point is typical: tens of thousands of stores worldwide consume massive volumes of perishables—including milk, syrups, and coffee beans—every day. Manual inventory checks are time-consuming and error-prone: staff may overlook a carton of milk, causing stockouts during peak hours; or over-report inventory, leading to spoilage and waste. Starbucks’ original vision was simple: staff would snap photos of shelves using tablets, and the algorithm would automatically identify and tally items, syncing data instantly to backend systems. That sounded appealing indeed—saving labor, reducing errors, and optimizing supply chains. Yet real-world “milk inventory” is far from the “standardized objects” seen through algorithmic eyes. Milk cartons may be crushed or deformed; syrup bottles may have worn or faded labels; shelf layouts vary dramatically across stores—details easily overlooked in lab-tested datasets, yet each one a potential “fatal trap” for AI in actual stores. ## II. Reality Is Harsh: What Exactly Went Wrong in Those Nine Months? According to internal employee feedback, the system’s errors were astonishingly diverse: mistaking empty milk cartons for full ones—causing the system to show sufficient stock while shelves were actually bare; interpreting shadows under syrup bottles as “extra inventory,” resulting in wasteful over-ordering; and even misidentifying cleaning tools on shelves as “new products” due to lighting changes. These elementary mistakes did not boost efficiency—in fact, they forced staff to spend even more time “correcting AI outputs”: what used to take minutes manually now became a vicious cycle of “take photo → correct errors → manually adjust entries.” A deeper issue lies in the AI system’s lack of “contextual semantic understanding.” For instance, staff might temporarily place a case of backup milk on the floor; the AI registers it as “inventory,” though those cartons haven’t yet been unpacked and entered into stock. Or, when promotional displays temporarily occupy shelf space, the AI conflates them with regular inventory items. Such common-sense distinctions—immediately obvious to humans—are insurmountable cognitive gaps for AI. ## III. Starbucks Is Neither the First Nor the Last Starbucks’ fiasco is no isolated incident. In recent years, Amazon scaled back its “unmanned store” concept after persistent misidentification issues; Walmart abandoned its AI-powered shelf-scanning system due to insufficient accuracy; and a domestic convenience store chain’s AI-driven forecast for “oden” sales led to severe stockouts—highlighting how “technology–market mismatch” has become an industry-wide problem. The root causes boil down to two key points: - **The contradiction between data quality and real-world scenario complexity**: No matter how pristine the labeled training data is in the lab, it cannot replicate unpredictable in-store events—such as baristas spilling coffee, customers moving items casually, or aging equipment producing blurry images.

Starbucks Scraps AI Inventory Tool Across North America

May 21, 2026

Starbucks has discontinued its AI-powered inventory app across North America due to miscounting issues with beverages like milk and syrups, opting instead for standardized counting methods to improve supply chain consistency.

reuters

2026-04-10

Starbucks UK Retail Arm Receives £13.7m Tax Credit Despite Sales Increase

2026-04-28

Starbucks Raises Forecasts After Strong Quarter, Shares Jump 5%

2026-04-29

Starbucks Shares Rise as Turnaround Efforts Start Showing Results

2026-05-20

US Group Says Starbucks Plastic Cups Marked Recyclable End Up in Trash

2026-05-21

Report: Starbucks 'Widely Recyclable' Plastic Cups Found in Landfills