NCR Ecosystem — Context for VARs

Understanding where NCR sits — and what it can't do — is the foundation of the Canary co-sell story.

What NCR is

A 140-year-old company that owns the physical retail transaction layer for approximately 80,000 store locations globally. The largest POS software supplier in the world. Currently in a painful transition from hardware company to software platform — shedding assets, cutting headcount, and chasing cloud margins while the market moves underneath them.

Key facts: - ~11% global POS market share - Split from NCR Atleos (ATMs) in October 2023 - Revenue $2.69B in 2025, declining — 2026 guidance $2.21–2.33B (down 13–18%) - Stock down 40% since December 2025 - CEO Jim Kelly (ex-EVO Payments, ex-Global Payments): financial turnaround focus, not retail AI - Jan 2026 NRF launch: "Voyix Commerce Platform" — microservices, AI analytics, all proprietary, no MCP

The architecture problem

NCR Counterpoint was built for a world where stores close at night and reconcile their numbers at end-of-day. That is still the architecture:

• Inventory reconciliation: batch — end-of-day or scheduled interval
• Reporting: batch — snapshots, not live data
• Multi-store sync: polling and batch replication — not event-driven
• Loyalty updates: post-transaction — not in-flight

MCP requires millisecond responses. NCR's best honest answer is "near real-time." Agents don't do "near."

The Voyix Commerce Platform launched in 2026 is microservices on top of legacy data models. The plumbing changed. The data architecture did not. Real-time event streaming is not in the public roadmap.

What NCR cannot build

NCR cannot: - Expose an MCP endpoint that customer agents can query - Enable touchless agent-authorised transactions - Run ALX/VSM — a store-side AI agent that negotiates in real time with a customer's Claude - Respond to customer agent queries in milliseconds at scale

And they won't build this soon: leadership is financial (not AI-native), the engineering challenge is structural (not a product cycle fix), and their VAR channel model means they'd have to cannibalise the partners who reach mid-market retail.

Why this is good news for VARs

NCR cannot control what VARs build on top of Counterpoint.

A VAR that adds an MCP layer — that builds ALX — owns the agent interface between NCR's platform and every customer's Claude. NCR cannot easily replicate this without cannibalising the channel they depend on to reach mid-market retail.

The VAR owns Layer 4. NCR owns Layers 1–3. Both survive. The retailer wins. See Architecture.

The window

NCR is in a financial transition, with leadership focused on cash flow and no public retail-AI roadmap. The MCP layer in the NCR VAR channel is the place where the agent interface gets built. Shopify launched MCP in Summer 2025; Walmart's Sparky launched the same year. The enterprise reference points exist. Physical retail POS is the next surface where MCP adoption applies.

Competitive context

The competitive table reflects MCP endpoint posture as of April 2026 across the surveyed tiers. Physical retail POS adoption of MCP is early.

Market size