After discussing the issue with @stefanbenten, our current best theory is that the number of concurrent DNS lookups required by the uplink during the upload is too high for the router you are using. For a single segment, we open 95 storagenode connections at once, and you might get disconnected from wifi if your router does not support this many DNS lookups.
On our end, we are working on a fix where the satellite returns the IP addresses for the storagenodes directly, rather than requiring the uplink to find these through the DNS.
To figure out if this might be the issue you are running into, please do the following:
In a separate terminal, run $ ping 8.8.8.8, then attempt your upload. Monitor the output of the ping command. If it spikes during the upload, it could be a hint that the concurrent DNS lookups are the issue.