$PENDLE killing vePENDLE for liquid sPENDLE is the most honest admission of tokenomics failure I’ve seen and everyone should study this
60x revenue growth. only 20% of supply engaged with vePENDLE.
lowest participation rate among ALL veToken models.
60% of pools unprofitable despite overall positive fee efficiency
that’s NOT success that’s design failure masked by a few whale pools carrying entire protocol
the problems @pendle_fi admits:
- multi-year locks destroyed capital efficiency
- weekly voting required expert-level defi knowledge
- rewards concentrated to tiny fraction who could navigate complexity
- vote-based emissions created terrible allocation (60% of pools bleeding value)
- composability killed by non-transferable locks
these are foundational flaws that made participation irrational for 80% of holders
- sPENDLE replacement: 14-day withdrawal. liquid composable fungible. 80% protocol revenue → PENDLE buybacks → distributed to stakers. algorithmic emissions cutting 30% waste
the loyalty bonus (4x boost for existing vePENDLE based on remaining lock) is damage control. don’t want current holders nuking token when locks become worthless
but here’s the real lesson:
- pendle generated $37M revenue in 2025 while only 20% of token holders participated.
- imagine if 80% participated but protocol only made $10M. everyone would call it failure
- instead they had revenue success despite tokenomics failure. proves product > token model. yield trading had PMF.
but why did it take years?
@CurveFinance established veToken model 2020. pendle launched vePENDLE knowing the limitations. still chose it. now abandoning it
but will sPENDLE actually increase participation from 20% to 60%+? depends if people actually want governance or if they just want yield
- if they just want yield (most do) then sPENDLE staking for buyback rewards without weekly voting works. set and forget
- if protocol still needs active governance then simplifying to “only vote when PPP active” might work. but most holders won’t care about governance even with low friction
the real test:
- does protocol revenue distribution to sPENDLE holders create better alignment than vePENDLE voting power
probably yes. direct revenue share = everyone wants protocol revenue to grow. voting power = whales extract value through governance
- killing LP reward boosting for new pools is significant. removing complexity. reducing edge for sophisticated users. democratizing access
- 2-year transition period for vePENDLE decay is long. creates two-class system. boosted legacy holders vs regular sPENDLE. might cause friction
overall this is what token evolution should look like: admit what’s broken. fix it completely. grandfather existing users. move forward
- most protocols would’ve kept vePENDLE forever rather than admit it failed. pendle deserves credit for honesty and willingness to pivot
my bet:
- participation increases meaningfully.
- liquid composable tokens with direct revenue share are obviously better than locked voting power for 95% of holders
the 5% who were sophisticated vePENDLE voters lose relative advantage. the 95% who were sitting out finally have reason to participate
that’s net positive for protocol even if top voters are unhappy
pendle revenue proves product works. sPENDLE fixes token. combination should compound
this should be case study for every protocol with failed veToken model. most won’t pivot because admitting failure is hard.
pendle did it anyway
respect





$37M revenue with only 20% veToken participation - that's the most honest failure metric I've seen in DeFi.
The honesty here matters. Most protocols double down on broken tokenomics. Pendle admitting that multi-year locks + weekly voting + expert-level complexity = participation collapse takes courage. The 4x legacy boost protecting existing holders while pivoting is smart design.
The core insight resonates beyond DeFi: product-market fit beats elegant system design. 60% of pools operating unprofitably despite protocol-level efficiency shows how distribution of value creation matters as much as aggregate generation.
I've been thinking about similar governance-vs-usability tradeoffs in AI agents: https://thoughts.jock.pl/p/ai-agent-comparison-claude-manus-zapier-real-testing-results-2025