yes, and it looks like a priming mistake by seeedstudio that is easily fixable.
(the 10 we bought last year all show the same broken behaviour.)
what’s wrong: they use an smsc95xx usb-to-ethernet chip, which has its own eeprom - but that eeprom ships EMPTY, for some stupid reason. the linux driver looks in the device tree for a mac address to set (uboot doesn’t pass one or not in a suitable spot), or in the chip’s eeprom (which is blank) and then falls back to a random mac.
solution: install ethtool and set the driver’s eeprom to something suitable.
for our boards we’ve fixed that this way:
ethtool -E eth0 magic 0x9500 offset 1 length 6 < /proc/device-tree/ocp/ethernet@4a100000/slave@4a100200/mac-address
this one-liner takes the mac address associated with the unused internal-to-the-core 10/100 ethernet and bonks it into the smsc’s eeprom. next boot you’ll see that same mac for your ethernet, and it’ll stay that way. (the manufacturer prefix is be 98:5D:AD, which belongs to TI).
as an aside, in the primary beagle eeprom at offset 60 you’ll also find a mac address with a seeedstudio-prefix, which nobody and nothing uses (that i know of) as that is a reserved block according to beaglebone docs. for our 10 bbgg boards that mac address is useless because they burned the same one on all boards,
see hexdump -C -n 4096 /sys/bus/i2c/devices/0-0050/eeprom which likely will show an identical ‘2CF7F1060001’ address.