What Retail AI and Compute Infrastructure Actually Looks Like in 2026

Old Hardware Stores Become New Again: Key Copying and Paint Mixing Systems

Let us give another really concrete example of how compute is coming to retail. An old-fashioned key copying rig used to require someone who knew what kind of key needed to be copied. Then, someone would have to manually grind it themselves. Nowadays, you have a computer that scans the key, knows which blank to pull, and then goes in and copies it with automated grinding. If you want to make multiple keys, the machine does it.

Another example that used to be somewhat of a skill and art form would be mixing paint. You had to know how to mix paint and get just the right tint. Nowadays, you can scan things, put them in a computer, and then the computer controls the mixing of the paint so you get much better accuracy, particularly if you do not have a ton of skill in mixing paints.

These systems are essentially specialized computers running custom software controlling mechanical hardware. The key copying machine has motors, sensors, and positioning systems all managed by embedded compute, then some sort of quality control mechanism. Paint mixing systems have a similar architecture with precision dispensing controlled by software. In any case, these are good examples of how even simple manual processes end up getting automated, again consuming more compute resources.

Customer Service Kiosks and Online Integration

In retail, there are a lot of other things you have to do beyond just selling products. With customer service, you may have folks looking up pricing or inventory online. Maybe you need to recommend a product that is not in stock today, but could be ordered and brought to the store later. Perhaps an employee needs to help a customer with a price match or a return. We often walk by these kiosks without giving them much thought.

Saying that retail has a lot of compute, you can see it all over the place once you start looking for it. These kiosks run full operating systems, connect to backend inventory and ordering systems, and provide customer-facing interfaces that need to be reliable and responsive. They can now also serve recommendations, so associates can offer alternatives or other products that are often used together. We were talking about keys and paint, but all of this is happening just a few meters away.

Spill Detection Algorithms

Say you have a needlessly careless customer (in this case, cameraman Sam) who decides to leave their drink in the middle of the floor. Inevitably, this is going to get kicked over, and there will be a spill. As a retailer, you would want to know if there is a spill in your establishment so you can dispatch someone to address it, clean it up, and make sure everything is safe. These days, applications can see safety hazards, even if employees are wearing protective gear.

Like so much of what we have seen, this requires computer vision models trained to identify liquid on floors, integration with alerting systems, and camera coverage in areas where spills are likely to occur.

Temperature Monitoring for Refrigerated Goods

If you have been watching STH for a long time, you know that I am a big barbecue and grill guy. While we are in a hardware store, they are also selling steaks to go with the grills they sell. Retailers selling food need to monitor that the food is kept at cold temperatures so it is safe to eat. The store needs to continuously monitor and track that. These days, food often has its own tracking systems.

Temperature monitoring infrastructure includes sensors in refrigerated cases, data logging systems, alerting when temperatures drift outside acceptable ranges, and integration with HVAC or refrigeration maintenance systems. This is IoT infrastructure running alongside all the other compute systems we have discussed. The reason that these steaks are there is really as an add-on for the grills and grill accessories, which is a use-case for recommendation systems.

Of course, shipping, receiving, and managing inventory is another area that is getting impacted.