So you need a primary chain (14 days) + BCJ chain (14+4+6+10)
You need round about
9TB (14 days) on your primary backup sitehttp://rps.dewin.me/?m=1&s=6700&r=14&c= ... y=1&dg=1&e
43TB (14+4+6+10) on your secondary offsitehttp://rps.dewin.me/?m=3&s=6700&r=14&c= ... 4,6,0,10&e
or 2x 52TB
With actual Version v9, and as you are a HPE shop, Usage of an HP Apollo (everything in one box) or DL380 with disk expansion would fit your need in case of performance and sizing.
As most of the data will be added over time, you can start with a smaller configuration and add capacity or an additional server with discspace over time.
Use Standard Settings for Veeam Dedup and compression.
This would allow later to use it with Win2016 ReFS Block Cloning and Veeam v9.5 which will reduce IO for Merge Processes dramatically but as well it uses the block cloning technology to reduce the Backup Copy Job GFS chains.
Maybe 2x (one for primary and one for secondary) HPE Apollo (or DL380) with ~25TB capacity and you need additional capacity (or Server) in 3-5 years?
When you go with this.... Use good Battery backed Raid controllers, add 2 SSDs and enable smart cache. Use Raid60 with 6+2 configuration.
Dedup Appliances will reduce performance and complexity dramatically and they are not needed at this data amount.
Maybe an option would be to use another server vendor as secondary with similar setup, just for safety. (A mistake or bug would not hit both).