Supply chain AI

Agentic AI replaces predictive AI as the supply-chain category-definer

Supply chain AI through 2024-2025 was predominantly predictive: forecast demand, anticipate disruptions, recommend responses. The 2026 inflection is agentic - AI agents that execute decisions across procurement, logistics, finance, and customer-service workflows without human approval at each step. Palantir's AIP plus its AI Hivemind orchestrator are the canonical example: a disruption triggers a coordinated agent swarm that updates logistics routing, finance accruals, and customer comms simultaneously, with Warp Speed packaging the pattern for manufacturing and MIB supply chains. Kinaxis (Maestro Agents and Maestro Agent Studio), o9, Blue Yonder, Project44, and FourKites have all shipped agentic modules through 2025-2026. The category test for 2026-2027 is whether agentic actions move from low-stakes (ETA updates, routine reorder) to high-stakes (multi-million-dollar contract renegotiation, inventory rebalancing across regions, sourcing-decision pivots).

State of the art (2026)

The 2026 state of the art is agentic, not predictive: AI that executes procurement, routing and rebalancing decisions rather than merely forecasting them. Palantir is the breakout platform - AIP Agent Studio plus the AI Hivemind orchestrator turn Foundry into an execution layer, and the $448M US Navy ShipOS contract (December 2025) pushes it into defence shipbuilding supply chains. Incumbents are retrofitting agents fast: SAP Joule, Oracle Fusion, Blue Yonder, o9 and Kinaxis Maestro. The category splits in two - real-time visibility (Project44, FourKites) commoditises while planning and autonomous execution stays premium. The open question is whether agents move from low-stakes reorders to high-stakes, multi-million-dollar decisions without human sign-off.

Palantir is the breakout enterprise platform - and now in defence supply chains

Palantir Foundry plus AIP have become the platform incumbent across the largest enterprise supply-chain deployments: Airbus (cross-tier microchip-delay visibility), Walmart, Amazon, ExxonMobil, Cardinal Health, Wendy's (inventory prediction). The agentic launches - AIP, AI Hivemind multi-agent orchestration, and the Warp Speed manufacturing OS (Lear, Cyberlux, Warp Speed for Warships) - convert Palantir from a data-integration platform into an operating system for autonomous supply-chain decisions. The December 2025 $448M US Navy ShipOS contract extends the pattern into defence supply chains - submarine programs at General Dynamics Electric Boat and HII Newport News now aggregate supplier and parts data on Palantir, initially across two shipbuilders, three shipyards, and 100 suppliers. The breakout question is whether classic SCM incumbents (SAP, Oracle, Blue Yonder, o9, Kinaxis) can retrofit agentic functionality fast enough or whether Palantir absorbs the top tier of the planning and execution market.

Visibility platforms commoditise while planning platforms premium up

The supply-chain AI category splits structurally into two layers. The visibility layer - real-time shipment tracking, multi-tier supplier monitoring, port-condition data - is commoditising as Project44, FourKites, and similar platforms reach broad enterprise adoption and as the underlying telematics data becomes table stakes. The planning and execution layer - demand planning, inventory optimisation, supplier-risk decisioning, autonomous procurement - is structurally premium and is where Palantir, Blue Yonder, o9, and Kinaxis compete on agent depth and integration. Expect visibility ARPU to compress through 2027 while planning-and-execution ARPU expands, particularly at the Fortune 500 tier where agentic decisioning has the most measurable ROI.