Following on from @'s earlier on Excel to Cloud migrations, a different possibility is also worth considering. David's approach is valid and leverages BI technologies to bootstrap a more powerful cloud enabled path. This is like turbocharging the existing design as Excel is seen naturally as part of the same ecosystem as SQL Server and Power BI.
The different possibility comes from skipping out of the BI ecosystem and landing heavily in the Big Data / Data Science realm. In order to do this, we should take the contents of the Excel file and drop the values not into a BI database but instead into a flat file and enable the downstream consumers from Big Data technologies.
As such, Parq, the command line tool that we talk about in relation to enables friction-free export of tables within Excel directly into parquet. This will then allow downstream data science to take place over the data held in Excel.
Watch out on the parquet dotnet channel for more information about this as it matures, and consider the following architectural design: