Context: what Compound was when we started (mid-2023)
When AlphaGrowth began working with Compound, V3 had just launched and Compound Labs had relinquished control of the protocol due to legal pressures in the US. Compound V3 was effectively an orphan protocol: no active development, no marketing, no partnerships. Meanwhile the dominant market narratives - Ethereum L2s, LRTs, BTC LSTs - were being captured by aggressive competitors. Aave and Morpho were eating market share, and Compound had no new chain or market launches in the pipeline.
That is the baseline against which our tenure should be measured: not "did every market thrive forever," but "what happened to a protocol nobody was managing, once someone started managing it." During our tenure, Compound’s total TVL grew from $1.41B to $2.10B (+49%) - a figure the analysis itself confirms.
Q2 - What supported the TVL where it held?
COMP emissions. COMP emissions are set by the protocol's risk manager (Gauntlet), not by AlphaGrowth, and they exist precisely to stimulate markets where interest-rate-curve mechanics alone can't compete. Treating them as a hidden subsidy that discounts our work gets the causality backwards: emissions are a standard, governance-controlled tool that every major lending protocol uses.
OKX Earn. We consider this one of our clearest wins, not a caveat. AlphaGrowth was among the first - possibly the first - in the industry to reach exchange DeFi-wallet users by incentivizing with ARB tokens. The ARB incentives let Compound jump a months-long integration queue at OKX. Exchanges hold the highest concentration of crypto users and are most users’ first step into the space; hand-holding them into DeFi through exchange partnerships has since become a mainstream strategy - see Morpho × Coinbase (morpho.org/coinbase-loans) and Aave × Binance on Plasma, which attracted over $500M in deposits. We specifically chose OKX because it charged no distribution fees and took no rewards for itself - 100% of incentives went to end users.
The five large addresses. Wallets of size control the majority of deposits on Compound and on every lending market - concentration is the norm in DeFi lending, not an anomaly of our cohort. More to the point: OKX DeFi Earn and OKX CeFi Earn are today major depositors across Compound markets. The partnership we built during the program is still returning sticky TVL after the incentives ended. That is what durable growth infrastructure looks like.
Q3 - Did incentive-free base APY become attractive?
This question deserves a technical answer, because raw base APY is the wrong yardstick for a single lending market.
In Compound V3-style markets, interest rate curves and final rates are set by risk managers (Gauntlet, LlamaRisk). By design, even at optimal ~90% utilization, the supply rate on a single market lands around 3–4%. This is not a flaw - it’s interest-rate-curve design that keeps DeFi lending competitive with real-world money markets. The analysis’s own benchmark table proves the point: Compound-Arb (2.63%), Aave-Arb (2.54%), Compound-Eth (3.22%), Aave-Eth (3.27%). No single-protocol lending market anywhere offers a standing "yield premium" - convergence to the benchmark is the healthy end state. The market we built functions today at market-standard rates holding $64.9M, as the analysis fairly acknowledges.
Base APY data: DefiLlama yields API (compound-v3, aave-v3 per-chain), as compiled in the track record analysis, pulled 2026-07-10.
A more potent metric for market success is utilization - how much of the base asset is actually borrowed. During AlphaGrowth’s tenure, the USDC, USDT and ETH markets continuously ran at 87–90% utilization, guaranteeing the best achievable rates. Compound markets today hover around 78%, producing suboptimal rates.
Interest rate model: Compound V3 USDT mainnet, live data pulled 2026-07-11 (v3-api.compound.finance; same figures visible at app.compound.xyz/markets/usdt-mainnet). At today’s 78% utilization the market pays 2.80% earn APR; at the 87–90% utilization maintained during AG’s tenure, rates sit at the kink - the optimal point of the curve.
Where does genuinely attractive organic APR come from? From stacking ecosystem apps into structured products: deposit USD → mint a CDP stablecoin → supply it as collateral on a lending market → borrow stablecoin → earn yield, as long as CDP APR exceeds the borrow rate. You can see this live with sFRAX collateral in the USDT mainnet market, utilized at 100% (app.compound.xyz/markets/usdt-mainnet). LST/LRT looping on the ETH market worked the same way - net APRs above 10% at its peak.
This is also why the distinction between protocol-level and chain-level engagement matters so much for PRIME. On Compound, we managed one protocol’s partnerships and variables. Chain-level management - which is what Cardano PRIME is - permits far greater creativity in composing structured products across the whole ecosystem to raise organic APR, with incentives acting as added firepower rather than the whole engine.
Q1 & Q4 - Did the flagship TVL stay, and where are all the markets now?
These two questions share one answer, so we address them together.
The Arbitrum campaign was a success by the measure that mattered at the time: it made Compound competitive on a chain where newer protocols with yield-farming flywheels (Morpho) and entrenched incumbents (Aave) were otherwise winning by default. The analysis’s own difference-in-differences measurement shows a genuine, controls-beating effect for 12–18 months (index 179–230 vs. 100 at grant end). That is what an incentive program is designed to buy: a window of competitiveness in which to build integrations and partnerships.
On address-count retention (0.8%): as the analysis fairly notes, this is the strictest possible metric. Large accounts rotate addresses, and TVL-weighted retention tells a different story. Capital in DeFi is continuously in flow; the relevant question is whether the partnerships and integrations built during the program continue to route capital to the protocol (see Q2).
What happened after the 12–18 month window comes down to four forces, and only one of them was within any growth program’s control:
- 1 Broad market decline - geopolitical shocks, BTC drawdowns, a continuous string of hacks (including this year’s LST meltdown and its lending contagion).
- 2 The structural failure of Ethereum L2s. 2026 marked a decisive shift of users and TVL away from L2s. The Dencun upgrade slashed Ethereum L1 transaction costs, making the L2 "cheap fees" pitch less compelling, if not redundant. Some L2s could not maintain their DeFi ecosystems; others never reached meaningful scale. Capital migrated back to Ethereum L1 or left DeFi entirely. Out of respect for the teams involved we won’t single out specific chains, but the pattern is broad: the trajectory of Compound’s deployments on these chains mirrored the general DeFi trajectory of the chains themselves. Notably, Aave and Morpho show the same trajectory: they too lost the majority of their TVL on L2 deployments. The 70–100% drawdowns in the referenced table are chain-beta, not program-specific decay.
- 3 Change of guard at Compound. When we handed management to the newly formed Compound Foundation, it shifted direction entirely - dedicating $20M to building a new product (Compound V4). There are currently no active partnerships or growth commitments at Compound. The post-handover TVL decline reflects that vacuum, not the quality of the growth that preceded it.
- 4 Capital velocity. TVL left unmanaged atrophies. Capital continuously migrates to actively managed opportunities - which is precisely why it flowed to Morpho (now launching payment and Coinbase integrations) and Aave after active management of Compound stopped. This is an argument for sustained growth programs, not against them.
The drawdown table makes this point on its own: markets that predate our involvement (Base −75%, Polygon −93%) and even Compound’s organic Ethereum home market (−52%) fell alongside the program markets. The decline is chain- and market-beta, not a program signature.
The counterexample proves the rule: the USDT mainnet market, which AlphaGrowth brought to Compound, has more or less maintained its form and TVL - $192M supplied at 78% utilization today (v3-api.compound.finance, 2026-07-11; verifiable on DefiLlama). Where the chain underneath held, the market we launched held.
Drawdown data: DefiLlama protocol API (compound-v3 per-chain TVL), as compiled in the track record analysis, pulled 2026-07-10; USDT mainnet market supply from v3-api.compound.finance, pulled 2026-07-11.
Q6 - How does PRIME’s KPI treat a trajectory shaped like this one? TVL atrophy without management
The honest answer to Q6 starts with naming what the flagship curve actually is: TVL atrophy without management. TVL is not a wind-up toy that keeps spinning after the operator walks away - it is a stock of capital that stays only as long as someone is actively giving it reasons to stay. The Arbitrum curve reverted after month 18 because, at precisely that point, active management of Compound ended: our program concluded, the Compound Foundation pivoted its entire budget ($20M) to building V4, and growth work stopped.
So the question "would fees be earned on growth that later reverts?" has the causality inverted. The reversion is not a property of the growth; it is a property of what came after - a management vacuum. The same force explains why capital flowed from Compound to Morpho and Aave, the protocols still doing active partnership work.
PRIME is designed around exactly this lesson, in three layers:
- 1 The atrophy problem is attacked directly, not just measured. The incentive taper (months 7–12), cliff-vesting cohorts, and vault products exist to transition TVL onto organic APR before incentives stop - the managed hand-off the Compound program never got the chance to complete. Organic APR is named in the proposal as the actual objective, not a talking point.
- 2 The fee mechanics already price in persistence risk. Fees accrue quarterly with a 30-day hold at every measurement, a declining marginal rate, and a hard $4.64M cap with unearned reserve returning to the treasury. A spike-and-revert curve earns materially less than sustained growth. Alongside this, we report a 6-month rolling persistence metric so the DAO watches the same number we do.
- 3 The DAO holds the off-switch throughout. The Month-4 release gate holds ~75% of capital until real data exists, six return triggers can send funds back to the treasury, and the month-6 falsification trigger commits us to recommending restructuring or pause ourselves if qualifying growth is below $80M with incentives flowing.
The deepest implication of Q6, though, is one we agree with: TVL needs a permanent operator, not a 12-month visitor. That is the strongest argument for Cardano building a durable growth function - which is what PRIME is structured to seed, not substitute for.
Owning our failures
Fair analysis cuts both ways, so here is what didn’t work - by our own doing:
The BTC mainnet market. We launched a BTC base market betting on the Babylon-driven native-BTC-yield narrative. Babylon’s yield proved insufficient to attract capital market-wide, and we recognized this too late - after deployment.
The wstETH market. We launched it to compete with Aave as EigenLayer brought LRTs to market. A pivotal player (Instadapp) chose Aave over Compound on the strength of a deeper existing relationship, and Compound’s then three-year-old architecture wasn’t the most profitable venue for the strategy. The market never gained traction.
Ronin. We aimed to be the first DeFi protocol on the Axie Infinity gaming chain and open it to DeFi. Its user base - young, gaming-focused, largely in Southeast Asia - never generated the capital or momentum required.
We name these because a growth partner that only narrates wins is not one a DAO should trust. PRIME’s design - the Month-4 release gate, six return triggers, the performance-fee cap with unearned reserve returning to the treasury, rolling persistence reporting, and the month-6 falsification trigger - exists precisely because we’ve learned what fails, not just what works.
Sources referenced: the track record analysis (adatool.net), Dune cohort data, DefiLlama protocol & yields APIs (api.llama.fi, live pull 2026-07-11), Compound governance records, app.compound.xyz, morpho.org/coinbase-loans, Binance Plasma announcement. All chart data points are independently verifiable against these sources.
