Podflow AI by Virtuals (POD)

Ran the floor again. It got worse for shorts.

Dolphin did 90.8B tokens yesterday (!)

source: https://t.co/DO3GeZNEhF

At OpenRouter rates for its workhorse model (Qwen 3.6 35B, $0.15 in / $1.00 out), output-weighted for the datagen load: ~$67K/day. ~$24M/year. 100% routed to POD buybacks.

And that is single-model math. The live basket (GLM 5.2, Qwen 27B, Gemma 31B/26B) makes it conservative on the revenue side.

$24M/year of buybacks against a ~$20M cap. The buyback is bigger than the entire market cap. Buyback-to-emissions: 7.6x. Net supply shrinks as it scales.

Venice does ~100B/day at a $616M cap. Dolphin is at ~90% of that throughput for ~2.4% of the valuation, and it builds the uncensored models that power the Venice ecosystem. And the API is not even live yet.

*Note for the FDV/MC crowd: that ratio only matters when unlocked supply dumps on you. Here every emission is met with 7.6x its size in buybacks. Supply contracts as usage grows. The FDV is not sell pressure waiting to hit you, it is the targ