When AMD's biggest market was Litecoin (and derivatives like Dogecoin) mining and Nvidia's hardware was pants at mining, they initially couldn't increase production of the HD 7000 series quickly enough, so the initial glut of money went to scalpers. They responded by making huge volumes for the Rx 200 series, but shortly after it launched, Litecoin mining ASICs became available and GPU mining stopped being viable. That meant that:

• they'd spent lots of money manufacturing lots of GPUs.
• miners were selling used GPUs for a fraction of the retail cost while those cards were still the current generation.
• people didn't want to buy a new card for several times the price of the same card but used for a few months.
• retailers had to drop prices to keep selling new cards.
• wholesale prices had to drop to keep retailers stocking new cards.
• AMD weren't making any profit when they sold these cards.
• the RX 300 cards weren't compelling compared to a massively discounted RX 200 card, so they didn't sell in huge quantities or with good margins, either.

This wasn't the only time ATi/AMD took a calculated risk and it backfired horribly, so with their history of bad luck, chasing the AI bubble in any way that involves risk instead of just selling things for money might be a bad idea.