Retail Is Out-Investing Banks in AI — And the Gap Is Widening
We spend a lot of time talking about AI in banking. DBS deploying agentic assistants to 10 million customers. JPMorgan building LLM-powered research tools. The fintech narrative dominates the headlines.
But look at what happened in retail over the past weeks, and the story flips. The world’s biggest retailers aren’t adding AI features to existing operations. They’re rebuilding their entire companies around AI — at a scale that makes most bank AI budgets look conservative. And the ones doing it aren’t startups. They’re a century-old department store chain, the world’s largest employer, and a tech giant that owns the full stack from LLM to payments.
Here’s what Lotte, Walmart, and Alibaba are actually doing — and why it matters.
—
Lotte: The Department Store That Became an AI Company
In July, Lotte Department Store launched the AI Biz Center — the first AI agent in the Korean department store industry dedicated to partner brands. That came on the back of an Innovation Award at Strategy World 2026 in February, where Lotte was recognized as the first Strategy customer to run Auto 2.0 AI agents in live production. Not for a chatbot or a recommendation engine. For production results that would make a tech company jealous.
The BI Agent. Launched in May 2025 and now in full production, Lotte’s BI Agent doesn’t just analyze customer purchase data. It pulls in external factors — weather patterns, foot traffic, market conditions — and synthesizes everything into conversational, real-time insights. The result after the first month of operation: data analysis time cut by 70%, and complex in-depth analyses up more than 10%. Analytical processes that once took hours now take under a minute. This isn’t a dashboard. It’s an AI that understands the business context of a department store and tells you what to do about it.
The AI Biz Center. Launched July 22, this industry-first agent automates the vendor onboarding process. New brands seeking store space describe their brand, target customer, and product strengths conversationally; the AI generates a draft proposal in Lotte’s internal review format, incorporates uploaded brand materials automatically, and routes it to the right department. Proposal-to-review time: cut by up to 80%. Think about what that means for competitive advantage. While a competitor’s leasing team is still reading PDFs, Lotte’s AI has already drafted the response, matched the brand to the right floor, and flagged the opportunity for review. Speed to market for new retail concepts just became an AI problem.
The context around it. This is the third pillar of a deliberate build-out: an AI interpretation service for foreign customers on transparent displays in 2024, the “The Dustin” AI chatbot in the Lotte app at the end of 2025, the BI Agent for employees in 2025, and now the AI Biz Center for partners — what Lotte itself calls a comprehensive AI business ecosystem across customer, employee, and partner domains. This is part of Chairman Dong-bin Shin’s group-wide AI transformation push — and it’s a department store chain doing it, not a tech company.
The pattern here is important. Lotte isn’t buying AI and bolting it onto existing processes. They’re rebuilding the processes themselves around AI. The BI Agent replaces manual analysis. The AI Biz Center replaces manual vendor screening. Each deployment isn’t a feature — it’s the replacement of a human workflow with an AI-native one.
Source: Strategy World 2026 | Seoul Economic Daily / Asia Business Daily, July 2026
—
Walmart: 2 Million Employees, One AI Supply Chain
Walmart’s AI story is different from Lotte’s, but the scale is staggering. They’re not just deploying AI — they’re retraining the world’s largest private workforce to work alongside it.
The supply chain. Walmart’s supply chain technology team, led by SVP Indira Uppuluri, runs AI models and agents across its network. Instead of looking at one node at a time, associates use agents that show how resources are leveraged as a whole and how to fix bottlenecks. Digital twins — virtual replicas of stores and the logistics network — simulate how the network responds to facility closures, transport delays, or sudden demand shifts. In stores, the digital twin has predicted refrigeration failures up to two weeks in advance, auto-generating the work order before anything breaks. In distribution centers, AI scans 100% of conveyable cases for defects and an AI Pallet Builder optimizes every load. “Assortment, speed, and cost” — that’s the triad the AI is balancing, continuously.
The workforce. Walmart partners with OpenAI and Google Gemini on role-specific AI certifications delivered through Squiggly, its associate-facing platform. The Google certificate launched earlier this year; the OpenAI certificate followed in June, with the first graduate being a 43-year veteran distribution-center manager — a deliberate signal. Certifications are currently open to 1.6–1.7 million associates in the US and Canada, with Chief People Officer Donna Morris’s stated ambition to equip all 2.1 million employees with some level of AI skill over the coming years. This is the part most AI deployment stories miss. You can deploy the best AI supply chain in the world, but if your workforce doesn’t know how to work with it, you’ve built an expensive paperweight. Walmart is treating AI adoption as a workforce transformation problem, not just a technology problem.
The implication. Walmart employs 2.1 million people globally. If they successfully retrain even half of them to work alongside AI agents, they’ll have created the largest AI-augmented workforce on the planet. That’s a competitive moat that no amount of venture capital can replicate. You can copy Walmart’s AI models. You can’t copy 2 million AI-trained employees.
Source: CIO Dive / Retail Dive, July 2026 | Fortune, February 2026 | Modern Retail, June 2026
—
Alibaba: Own the LLM, Own the Stack, Own the Customer
This is where the story gets strategic. Alibaba isn’t buying AI from OpenAI or Google. They built Qwen — their own large language model family — and Qwen 3.8-Max, previewed in mid-July and fully launched this week, is a 2.4-trillion-parameter model Alibaba positions as second only to Anthropic’s Claude Fable 5 in reasoning and agentic tasks (independent benchmarks are still pending — treat the ranking as a claim, not a fact).
The commerce play. Alipay’s AI Pay processed more than 120 million AI-agent transactions in a single week in February 2026 — the world’s first AI agentic payment service to hit that volume. These were completed purchases: AI agents discovering, selecting, and paying without the user leaving the conversation. The Qwen App was the first platform to adopt Alipay’s Agentic Commerce Trust Protocol, launched in January, wiring conversational shopping directly into Taobao Instant Commerce and Alipay AI Pay. That’s not a pilot. That’s agentic commerce at consumer scale — and Alibaba built the AI, the protocol, and the payment rails themselves.
The luxury play. LVMH runs on Alibaba’s infrastructure in China. The partnership, dating to 2019 and extended for five years in 2024, has LVMH integrating Alibaba Cloud’s Qwen LLM and Model Studio (Bailian) for generative AI-powered luxury experiences. Around 30 LVMH Maisons are on Tmall Luxury Pavilion, using virtual try-ons, 3D displays, and livestreaming — all powered by Alibaba’s AI infrastructure.
LVMH deployed its own generative AI agent, MaIA, to some 40,000 employees for client advisors. But the infrastructure underneath in China — the LLM, the cloud, the commerce platform, the payments rails — is Alibaba’s. LVMH’s philosophy is “technology needs to be everywhere, but visible nowhere.” Each of their 75 Maisons has its own AI transformation plan. And in China, the engine powering it is Chinese.
The payments play. This is the part that should worry banks. Alibaba’s AI stack doesn’t stop at product discovery and recommendation. It extends through the entire transaction — from AI-powered search to AI-powered checkout to AI-powered payments via Alipay. When a consumer chats with Qwen to order, the AI handles discovery, comparison, purchase, and payment in a single flow. No bank. No third-party payment processor. Just Alibaba’s stack, end to end.
The strategic implication. Alibaba isn’t a retailer using AI. It’s an AI company that owns the retail stack — the model, the cloud, the marketplace, and the payments. Every transaction that flows through that stack makes the AI smarter, the recommendations better, and the moat deeper. The more LVMH and other brands build on Alibaba’s infrastructure, the harder it becomes to leave. This is platform lock-in at the AI level.
Source: Business Wire / Alipay, February 2026 | NielsenIQ, July 2026 | LVMH, May 2024 | Alibaba Qwen launch coverage, July–August 2026