I was following the Snowflake Q3 earnings presentation a few weeks ago and my jaw dropped, when I saw the following slide:
Are they really going after building a platform for warehouse-native apps? It seems quite simple at first, but disrupting app development by potentially introducing a hub-and-spoke architecture, would be mean an introduction of a fundamental re-architecture of how SaaS apps could work.
Today most apps work with a point-to-point data model, which means each software vendor controls some database in solitude which is the underlying foundation of the app. For example, an ERP system like SAP or Microsoft Business Central essentially has a large database with their data inside to run the app. Same for e.g. a CRM like Salesforce. Simplified, it looks like this (point-to-point):
Companies today use hundreds of SaaS apps, each running a database in solitude. This makes sense given the history of software infrastructure, however it gives huge leverage to the vendor e.g. an ERP system, because they fully control the data, that’s essentially the source of truth for your business.
There are multiple drivers for why vendors, owning their customer’s data, is ripe for disruption by players like Snowflake.
First is control. Controlling data access centrally allows for efficient and granular controls for apps - something that is top of mind for larger enterprises.
Second is having efficient legal processes. More and more regulation such as GDPR is coming around customer data, and companies will want to control data centrally to mitigate compliance risks and speed up legal processes.
Third, there’s cost. Owning your own app data, will change the power dynamics of SaaS vendors and customers which can more easily switch providers and with hosting the infrastructure on your own central data warehouse, can push prices down. How will that leave SaaS gross margins? Time will tell, but it could be a driver for more competition and lower margins.
Lastly, it would eliminate the pain of always missing data in the right system and having a spaghetti of integrations between them. With a central data store, the apps can query exactly what they need to run the app, within your set of permissions for what they can access.
The model would probably end up looking something like the following (hub-and-spoke):
The SaaS apps now query the central data store, and they can slice & dice the data for their app to run perfectly.
Back to Snowflake. When you think about it, this could be a huge opportunity for them to become “The central data store for all enterprise SaaS apps”. And it fits their target customer of large enterprises perfectly because they care a lot about data ownership due to the reasons above.
The reasons are many for why a modern data warehouse is the perfect tool for a central data store - it’s scalable, elastic, affordable and the data is yours. That said, technically there’s still some way to go, and I think there would have to be certain developer infrastructure like e.g. Supabase but on a warehouse, in order to see mass adoption of warehouse-native app architecture.
While it is definitely easier said than done to implement a warehouse-native / hub-and-spoke model, I do believe this is the way enterprise SaaS products will be built in the future. And we’re already seeing startups build products like this - a recent example being an email marketing SaaS app Castled built on top data warehouses like Snowflake. I would also expect to see a CRM system and billing system entirely built on a warehouse in the near future.
This re-architecture enabling warehouse-native apps, i.e. hub-and-spoke model, for enterprise software is definitely ambitious and will take time. However, Snowflake is in the perfect position to pull it off and build a new era of SaaS.