Six Retail IT Priorities To Keep Every Store Selling While You Modernize

Retail sales keep climbing in dollar terms, while margins stay thin. For example, Food retailers' average profit margin was 2.1% in 2025, according to FMI's The Food Retailing Industry Speaks 2026; about two cents on every dollar of sales. Meanwhile the store has become a small data center often with little or no IT to support it, and the AI the business is betting on increasingly needs to run at the shelf edge.

The takeaway: retail IT can meet this moment by simplifying, not just spending more. It can pursue a common platform and operating model across the store, the data center and the cloud, with a rollout that starts with one store. Here are six priorities we are seeing in the field.

Headline retail sales can look healthy while the aisles tell a different story. Sales are reported in dollars, not units, so revenue can rise while fewer items leave the shelf. The top line grows; the IT budget doesn't.

At the same time, the store has taken on more jobs. It sells, fulfills online orders, processes returns and controls shrink, all inside the same four walls. In a Coresight Research survey, US retail decision-makers put the cost of in-store inefficiencies at 6.4% of gross sales in 2026, up from 4.5% in 2024. That is roughly $196 billion.

Retailers are responding in two ways. Some invest to widen their advantage, and 93% believe AI adoption will give them an edge. Others simplify and cut IT costs to free up capital. Both paths lead to the same requirement: lower the cost to serve, reduce complexity and build capacity for AI.

1. Edge: run more in every store with a lean on-site footprint. A store now runs point of sale, payments, inventory, loss-prevention cameras, self-checkout vision, digital signage and sometimes GPUs, often with no IT staff on site. Zero-touch provisioning, central lifecycle management and applications designed to keep running when the network drops let a small team run hundreds of locations. 

2. Core: modernize the systems the business runs on. ERP, order management and loyalty databases still run the business. Consolidating compute, storage and virtualization on a common platform can help address infrastructure silos. In a published customer story, a home improvement retailer reported reducing capacity provisioning time from about three weeks to under an hour.

3. AI: bring AI to where the sales are. A 2026 Kyndryl report states that more than 80% of retail sales still occur in physical stores. Some self-checkout vision workloads need low-latency responses and may need to keep working through intermittent connectivity, so some inference can belong in the store.

4. Data: know what you have, where it is and who bought it. AI shopping assistants and marketplaces now place orders against retail systems directly. If an agent can't confirm what's on the shelf, the order can go to a competitor that can. Yet only 15% of retail leaders say they use their omnichannel systems to full potential.

5. Resilience: keep the store open, whatever happens. Hardware failures, storms, ransomware and peak-season outages all threaten the till. Microsegmentation, broad encryption and rehearsed failover turn recovery into a procedure. A Caribbean grocer moved to its secondary data center within three hours during a 2022 hurricane and ran there for eight days.

6. Hybrid: run the store, data center and cloud as one estate. For most retailers, hybrid is increasingly the destination, not a phase. Burst to the cloud for peaks, keep steady workloads where they run most cost-efficiently, and move many workloads between them without refactoring. That matters when roughly three in ten cloud dollars are wasted.

Retail can prove a new platform in one real store and limit the exposure to one building. Run it beside what's there, extend to a region, then bring the data center and cloud onto the same operating model. For many retailers, a spring pilot keeps the work clear of the holiday change freeze.

1. Prove it in one store, running the new platform beside the existing one with a real workload.
2. Extend to a region. Remote, repeatable provisioning makes each next store quicker to bring online.
3. Bring the data center and cloud onto the same operating model, so workloads can move between them.
4. Work the calendar. Pilot in spring and roll out regionally in summer, clear of the holiday change freeze.

What This Looks Like With Nutanix

The six priorities share one answer: run every location on the same platform, with the same tools and the same team. With Nutanix, that means:

• One platform from shelf to cloud. The Nutanix Cloud Platform runs virtual machines and containers in a one-node store, a data center or a public cloud, managed from a single console (Prism Central).
• Stores that need fewer visits. Zero-touch provisioning and fleet-wide lifecycle management can help reduce truck rolls. See the Enterprise Edge AI design.
• An alternative virtualization approach. With AHV and Nutanix migration tooling, Nutanix gives customers evaluating VMware environments an alternative virtualization platform.
• AI placed by workload requirements. Nutanix Enterprise AI can support selected inference workloads across store, data center and supported public cloud deployments.
• Data and resilience. Nutanix Database Service helps automate database lifecycle tasks for systems like POS and orders, and Nutanix disaster recovery capabilities let teams test failover on a schedule.

Keep Every Store Selling, Starting With One

Retail IT is unlikely to get a bigger budget for AI and the reinvented store. What it can pursue is a more consistent operating model across the estate, and a rollout that proves itself one store at a time.

Try the platform in your browser or explore Nutanix for retail.