@kami all great replies so far, and as others have pointed out it mostly boils down to a truly NVR-less solution, and it's 100% browser-native so there's no special software or browser plug-ins or controls to add. It's any browser on any device, anywhere, anytime and you've got full access and control. It's just a camera, a license, and you're done. Add in the other items like it's very bandwidth conscious, and video is encrypted at rest, encrypted in transit, with encrypted management traffic so it's also super-secure. As others have pointed out, eliminating the NVR removes any scalability concerns as well as a common attack vector into many infrastructures.
Apart from the MV solution itself, don't overlook all the benefits of using it within the Meraki Dashboard environment. That alone makes MV better by default. Things like RBAC and 2 factor authentication, local time zones, offline device alerting, simple and consistent firmware management, connectivity history, and multiple self-maintaining spreadsheets for things like inventory, change log, licensing, etc all built into the Dashboard and there's no extra effort or cost to manage any of those things.
Sometimes just one or two of these things are enough for some customers to adopt MV, but all of them taken in combination is very compelling and one of the reasons MV has become one of Meraki's fastest-ramping products ever.
One more item I'll add that helps differentiate Meraki is the advanced analytics and the MV Sense API. This is true machine learning at work. Meraki MV cameras constantly train and re-train a neural network that leverages every camera as a sensor, and the more customers who deploy MV, the smarter the system gets (for example about what is and is not human motion) and the more customers benefit from one another.
In addition to finding out how many people were in a certain place in the past or the present moment, the advanced analytics can be coupled with MV Sense and a sub-second MQTT feed to meet various use cases. For example, human motion detection could trigger smart lighting, or you can find out how many people are still being detected in a building that was evacuated, or trigger a media display in a retail store when there's more than 5 people nearby or analyze wait times at the cash registers, or in a manufacturing plant it can send an alert and shut off a dangerous piece of machinery if there's not at least 2 people in the room with it. Stuff like that, just a few examples.
The analytics are already pretty impressive and of course more will come in time. MV Sense allows customers to access that person-detection data (time, location, count) through a set of API endpoints to build new tools of your own to meet your specific use cases.