Santiago Roel Santos’ recent piece “Why Tokens Can’t Compound” is one of the sharpest, most honest diagnoses of crypto tokenomics I’ve read.
The core argument is correct in its historical framing:
Traditional equity compounds because companies retain earnings, reinvest them, allocate capital, expand margins, and buy back shares. Most tokens were deliberately engineered not to do that — to avoid looking like unregistered securities. Fees get distributed, treasuries are often spent on grants or ops, and there is no enforceable mechanism for retained earnings or compounding growth engines.
But the article’s conclusion — that tokens structurally cannot compound and that all the value will flow to equity in Labs/companies building on the rails — is already being disproven in real time.
The Missing Piece: Revenue-Funded Buybacks + Burns Are the New Reinvestment Engine
In 2025 alone, protocols spent over $880M–$1.4B on token buybacks (CoinGecko data). Some of the largest and most credible programs:
Hyperliquid (HYPE) → $644M+ committed through Oct 2025, buying back ~16% of circulating supply. Rule-based, funded directly by trading fees, executed transparently. At one point the Assistance Fund had repurchased tokens worth nearly $1B at current prices.
Pump.fun (PUMP) → $138M–$233M spent, retiring ~3–6% of supply.
Jupiter (JUP) → $57M+, 50% of trading fees directed to buybacks + extended locks.
Optimism (OP) → Starting Feb 2026, 50% of Superchain sequencer revenue (~$9M annualized run-rate) goes to recurring OP buybacks. Tokens go to treasury (for potential future burn/stake/redistribution).
Others: Ethena, Helium (paused but ran one), Uniswap (large treasury burn of 100M UNI in Dec 2025), BNB (ongoing quarterly burns), etc.
This is not “cosmetic tokenomics.” When buybacks are:
- Funded by recurring protocol revenue (not treasury runway),
- Rule-based and transparent, and
- Net of emissions/unlocks (the real metric is “buyback coverage ratio” > 1),
they create the exact same economic effect as corporate share repurchases: supply reduction + value accretion to remaining holders.
Mathematically, it’s compounding via scarcity. If usage → fees → buybacks → lower floating supply, then the same (or growing) fee stream accrues to fewer tokens. That is compounding.
On-Chain Treasuries + DAO Reinvestment = Equity-Like Capital Allocation
Many protocols now run protocol-owned treasuries that actively reinvest: grants, R&D, ecosystem funds, acquisitions, liquidity provision, even starting new products.
veToken/lock-up models (Curve, Balancer, now many others) force long-term alignment and give voters boosted yield — effectively a compounding flywheel for committed capital.
Some DAOs have moved toward capped spending + excess revenue → buyback/burn rules, exactly like a dividend policy + buyback authorization.
The legal objection (“we can’t retain earnings without looking like equity”) is weakening fast. Regulatory clarity is arriving, and projects are simply shipping anyway. 2026 feels different.
Where the Original Thesis Still Holds (and Where It Doesn’t)
Santiago is right that:
- Most early token designs were garbage (infinite emissions, no revenue link, Labs capturing all the real equity upside).
- Pure “usage = value” narratives without any supply-side discipline have failed repeatedly.
- Staking yields on most L1s are just floating-rate coupons on usage/inflation, not true compounding.
But the counterpoint is already live: the best projects are retrofitting compounding mechanisms into tokens. Hyperliquid, Optimism Superchain, Pump.fun, Jupiter, and a growing cohort are showing that tokens can capture and compound protocol-level value when the tokenomics are designed (or redesigned) for it.
The Real Trade in 2026+
Bearish on “default” tokenomics — yes, 90%+ of tokens will continue to act like volatile, non-compounding coupons.
Bullish on tokens with credible, revenue-linked, net-supply-reducing mechanisms — these are starting to behave like equity with superpowers (composability, global instant settlement, permissionless access).
Still very bullish on equity in companies leveraging the rails (the original article’s conclusion) — but now we have a third bucket: well-engineered tokens that compound.
The internet didn’t make TCP/IP tokens rich. It made Google, Amazon, etc. rich. Crypto is doing the same — but this time some of the “picks and shovels” are issuing tokens that are learning how to compound like equity.
The design flaw was real. The adaptation is happening faster than most expected.
Tokens can compound. The ones that figure it out first will be the Berkshire Hathaways of the on-chain economy.
