Every neat little delivery date on a checkout page conceals an argument. The website says the blue jacket exists. A warehouse says it is in bin C-18. A store associate may have just sold the last one. A carrier has a cutoff in 11 minutes. Somewhere, software must decide whether the promise is real - and what sequence of people, shelves, robots and trucks can keep it.
Manhattan Associates sells that decision-making machinery. The Atlanta company builds systems for demand planning, inventory, order management, point of sale, warehouse work and transportation. Its customers are mostly the organizations with enough volume and complexity to make a spreadsheet feel like a practical joke: large retailers, manufacturers, wholesalers, distributors and logistics providers. More than 1,200 use its products.
The company is public, profitable and easy to miss. Shoppers rarely encounter its name. They encounter its consequences: a credible pickup time, a store that can ship an online order, a picker sent down the sensible aisle, or a load assembled without wasting half a trailer. Manhattan fits between the enterprise resource planning system that records the business and the physical operation that must make the record true.
An underwear problem with excellent timing
The origin story begins in 1990 with Jockey, the underwear maker, and five technologists working on an ERP installation. They noticed a gap: manufacturers could make the goods, but shipping them to major retailers meant complying with a thicket of labels, cartons and routing instructions. The team created a packaged Pick Ticket Management System, known as PkMS, to make those outbound details manageable and upgradeable.
They called the company Manhattan Associates after Manhattan Beach, California - not the borough 2,800 miles away. Alan Dabbiere became the founding chief executive; Deepak Raghavan the founding technology chief. Public filings also identify Deepak Rao and Ponnambalam Muthiah as founders. Jockey became customer number one in 1991 and remains a customer, a rare bit of corporate continuity that doubles as a lesson about this market. Supply-chain software is not casually replaced. It sits close to the cash register and the loading dock.
The company moved to Atlanta in 1995, went public in 1998 and widened its reach through acquisitions. Intrepa brought transportation management in 2000. Logistics.com followed in 2002, the same year Manhattan opened a development center in India. Evant added demand planning, forecasting and replenishment in 2005. By then, the software no longer merely printed the right label. It helped decide what should move, from where and when.
The platform bet
Manhattan Active, launched in 2017, is the center of the current business. It is a cloud-native, microservices-based platform on which the company's applications share data and services. "Evergreen" is the enterprise-software word Manhattan uses for continuous updates: customers receive new capability without waiting for a giant, episodic version upgrade.
The architecture is the competitive argument. A retailer can buy warehouse, transport or order software from several vendors and integrate them. It can use modules inside SAP or Oracle. It can choose specialists such as Blue Yonder, Infios, e2open or RELEX. Manhattan's pitch is that one operational platform reduces the seams where data ages, integrations break and two applications optimize against different assumptions.
Manhattan Active Warehouse Management directs inventory, labor, slotting, yard work and automation inside distribution centers. Transportation Management plans carriers, rates, routes, loads and fleets. Active Omni handles orders, delivery promises, fulfillment, self-service and customer engagement. Store applications connect point of sale with network inventory and ship-from-store work. Planning tools forecast demand, replenish stock and allocate products.
The practical benefit is not simply visibility. It is the ability to change the plan. If a storm closes a route, if an item sells faster than forecast, or if a warehouse shift is short-handed, the system can reconsider inventory, labor and transport together. The company maintains a research group of math and science specialists, including PhD-level experts, to build the optimization algorithms underneath those choices. It spent about $145 million on research and development in 2025, primarily in the United States and India.
“They don't just assist - they act.”Sanjeev Siotia, chief technology officer, on Manhattan's embedded AI agents
AI that has somewhere to work
In January 2026, Manhattan made its AI agents commercially available within Active applications. The distinction it stresses is location. These are not chat windows perched above an old data lake. They operate inside the software that already knows the order, task, worker and shipment. Interactive agents guide people; autonomous agents monitor conditions and can take or recommend corrective action.
A Store Associate Agent can summarize sales performance. A Contact Center Agent can assemble customer and order context. A Labor Agent can recommend how to deploy workers against remaining tasks. A Shipment Tracking Agent can spot trouble and propose compensation. Eaton, the power-management company, tested warehouse agents for wave coordination and labor planning. Agent Foundry gives customers a governed, no-code way to build or customize agents with natural language, APIs and guardrails.
That makes Manhattan's AI proposition more concrete than most. An agent is valuable not because it produces elegant prose, but because it knows which late shipment matters and what operational choices remain. The risk is equally concrete: software that can act needs permissions, audit trails, reliable context and clear human escalation. Manhattan's advantage is access to the workflow. Its burden is proving the automation deserves it.
How the money moves
Manhattan is becoming a subscription company without pretending that enterprise change installs itself. Cloud contracts typically run five years or longer. In 2025, cloud subscriptions produced $408.1 million, up 21 percent, but professional services remained the largest line at $503 million. Consultants plan implementations, move historical data, configure systems, train teams and guide upgrades. Most customers use at least some of this help.
Maintenance from older licensed installations contributed 12 percent, while new software licenses and resold hardware supplied the small remainder. The direction is plain: maintenance declines as customers move to Active, and recurring cloud revenue takes its place. By June 2026, contracted revenue not yet recognized - remaining performance obligations - had reached $2.47 billion. Q2 cloud subscription revenue rose 26 percent from a year earlier.
The model also explains why this category resists tidy SaaS comparisons. A warehouse migration can affect conveyors, radios, labor standards, carrier agreements and the final promise displayed online. That complexity lengthens sales cycles and makes services essential. It also creates durable relationships once the system works. The top five customers supplied only 10 percent of 2025 revenue, so the business is not balanced on one giant retailer.
Leadership changed in February 2025, when Eric Clark succeeded longtime chief executive Eddie Capel. Clark arrived after running NTT DATA North America; Capel, who had led Manhattan since 2013 and worked there since 2000, became executive chairman. The handoff paired an outside operator with institutional memory at board level. Across the company, more than 4,150 employees work on five continents. Its large India development operation is particularly important: Manhattan conducts most product research internally in the United States and India, arguing that keeping the work close preserves both domain knowledge and programming continuity. That is a sensible cultural preference for software whose strangest edge cases often live inside a customer's loading schedule, labor rule or returns process.
The market behind the market
Manhattan sits in a valuable middle territory. ERP vendors own broad corporate records. Automation companies own robots and machinery. Commerce platforms own storefronts. Carriers own the movement. Manhattan tries to orchestrate the decisions between them. Partnerships extend that reach: Google Cloud provides infrastructure and marketplace distribution; Shopify connects storefront demand to post-purchase optimization; consulting firms handle transformations; the Manhattan Automation Network brings robotics providers into the warehouse ecosystem.
Its customers illustrate the span. Pet Supplies Plus uses cloud warehouse technology across a network of more than 700 stores. KeHE has reported a 25 percent reduction in spoilage at distribution centers using Manhattan forecasting and replenishment. Pacsun used ship-from-store capabilities to keep selling inventory from closed locations during the pandemic and later implemented Active Point of Sale. L'Oréal, Foot Locker, Floor & Decor, Staples, Cardinal Health and ICA Sweden appear in public customer material.
Analyst recognition matters in this conservative market because buyers shortlist systems that will run for years. Manhattan lists 18 appearances as a leader in Gartner's warehouse-management evaluation through 2025, eight consecutive leadership placements for transportation through 2026, and six leadership placements in Forrester's order-management evaluations. It also became a leader in Forrester's 2024 point-of-service assessment. The counts are less interesting as trophies than as evidence of sustained category presence across several operational layers.
The company's latest chapter is a test of whether those layers can become one learning system. If Manhattan can make warehouse, order, store and transportation data useful to agents without making operations reckless, it moves from recording and optimizing work toward performing parts of it. The shopper may never know. Ideally, the jacket simply arrives when the page said it would.