SpiderOak One and Groups have no hard limit on the size of a file that you can back up, sync, share, or download. There are however practical limits that cause us to recommend that you limit yourself to individual files no larger than 10 GB or so for backup, and no larger than 3 GB or so if they are to be synchronized (that includes being placed in Hive), placed in a ShareRoom, or shared via a file sharing link.
For your protection, One and Groups encrypt on your computer all your files before backing them up or syncing them, and decrypts them on your computer immediately after downloading. Encrypting and decrypting files of even normal size requires considerable resources on your computer. The larger the file to be encrypted, the more likely it is that your computer will not be up to the task.
One and Groups are designed for backing up personal files, such as documents, music, photos, and videos, and is optimized for the typical sizes of such files. They are not designed to work with huge files such as typically found with ISOs, virtual machine images, encrypted volumes, and so forth. For more information on what kinds of files One and Groups are designed to back up, see What Data to Select for Backup.
ShareRooms are a special case. Your visitors will access the files via an HTTPS download, and HTTPS was not designed to deliver particularly large files. While you can place a huge file in a ShareRoom without error on your end, your visitors will likely experience timeouts in their web browser before the file is fully delivered.
Similarly, how long it takes for our servers to generate a file sharing link depends upon the size of the file, and a huge file might take so long that the application on your computer concludes that the process failed and show you an error message. Our advice is to not try to generate a link for files larger than One and Groups are intended for. However, a workaround is to attempt to create a file sharing link, and allow it fail. Wait one hour, then try again. In some cases the second attempt will succeed, using the cached results from the previous attempt.