A bonding curve is a smart contract that sells a token along a preset price formula: the more tokens people buy, the higher the next token costs. No order book, no liquidity pool, no market maker on day one — the formula is the market. It's how nearly every memecoin launchpad works, from Pump.fun and BONK.fun on Solana to Four.meme and Flap on BNB Chain to Pons on Robinhood.
Most explanations stop at "price goes up when people buy." That's the least interesting part. The interesting part is that every launchpad tunes its curve differently — where the price starts, how hard it bends, how much capital it takes to finish — and those choices decide what a $3,000 buy actually gets you on each venue. So instead of another diagram of a line going up, here is every major launchpad's actual curve on one chart, running the same math Sumo's launch wizard uses before a real launch. Drag to buy; then switch tabs and design a curve of your own:
How a bonding curve actually works
Under almost every launchpad curve is the same machine: a constant-product market, x·y=k — the Uniswap formula — with one twist. Instead of real deposited liquidity, the contract starts with virtual reserves: made-up balances that exist only to set the price.
Pump.fun's SOL curve starts with 30 virtual SOL against 1,073,000,000 virtual tokens. Nobody deposited that SOL; it's a constant in the program. But the pricing math treats it as real: your buy adds SOL to the virtual reserve and removes tokens, and the ratio of the two reserves is the price. Thirty SOL against ~1.07B tokens prices the token at about 28 nano-SOL — roughly a $2,000 starting market cap at $74 SOL, before anyone has bought anything.
That's the whole trick, and it's why launchpads exist:
- Launching costs almost nothing. A Uniswap-style pool needs real liquidity to set a price. Virtual reserves set a price with zero capital — which is why deploying on Pump.fun is free and launching a memecoin costs only whatever you choose to buy.
- The price path is deterministic. Two virtual reserve numbers fully determine what every SOL of buying does to the price. There is no liquidity to pull, thin out, or fake.
- Every buy has an exit. The curve is always willing to buy tokens back at the formula price. Sellers push the virtual reserves back down the same path buyers pushed them up.
The curve bends upward because each buy shrinks the token reserve that divides the quote reserve. Early on the token reserve is huge and buys barely move it; near the end of the curve the same buy moves the price several times as much. That convexity is visible in the chart above — and it's the mathematical reason bundled first-block buys are so much cheaper than buying the same supply later.
Why every launchpad's curve is different
Launchpads differentiate on exactly three numbers: where the curve starts, how much it takes to finish, and how much of the supply is sold along the way. Everything else is branding. At SOL $74, BNB $571, and ETH $1,915 (the tool fetches live prices, so its numbers will drift slightly from these):
- Pump.fun starts around a $2.1k market cap and graduates after about 85 SOL (~$6.3k) of net buys, selling 79.3% of supply. Its USDC curve is the same shape re-quoted: 4,292 virtual USDC, graduating after ~12,161 USDC.
- BONK.fun doesn't hardcode virtual reserves at all — Raydium LaunchLab derives them from an 85 SOL fundraising target, which lands within rounding of Pump.fun's curve. Same destination, opposite direction: Pump.fun picks reserves and gets a raise; LaunchLab picks a raise and computes reserves. (The "design your own curve" tab does exactly this derivation.)
- Four.meme starts near $3.3k and needs ~18 BNB (~$10.3k) to graduate; Flap starts slightly lower and finishes at ~16 BNB. Nearly identical constants, visibly different endings.
- Pons V2 on Robinhood runs the steepest mainstream curve: 1.68 virtual ETH against exactly 1B virtual tokens, graduating after just ~4.2 ETH — a curve built for a chain where blockspace is cheap and attention is short.
- America.Fun is the odd one out: a linear raise — 12,500 USD1 buys 800M tokens at a fixed rate rather than a compounding curve. On the chart it's the flat line that only lifts near the end, exactly what its DBC configuration is designed to do.
Same $3,000, five very different outcomes — which supply share you end up owning, and at what market cap your chart opens, is a venue decision as much as a budget decision. For the full launch-day economics (fees, bundles, dev buys, direct pool launches), run the launch cost simulator.
Graduation: where the curve ends
A bonding curve is scaffolding, not a home. When cumulative net buys hit the curve's target, the launchpad "graduates" the token: it takes the quote it collected plus the unsold supply, seeds a real liquidity pool on a DEX — PumpSwap for Pump.fun, Raydium for BONK.fun, PancakeSwap for Four.meme and Flap, a Pons V4 pool for Pons — and retires the curve. From that block on, the token trades against real reserves, LPs, and everyone else on the chain.
Graduation is also where the trading changes character. On the curve, the only participants are people interacting with the launchpad. On the DEX, you inherit arbitrage bots, snipers, and MEV — and the deterministic price path is gone. The trading work runs through both phases: on the curve, where your dev buy, bundle, and early supply management decide how the market opens, and after graduation, where most of the volume lives and market making, counter trading, and exits carry the token. A launch that treats the curve as a waiting room usually graduates into a chart it doesn't control.
Bonding curve vs. liquidity pool launch
Not every token starts on a curve. A direct pool launch — Uniswap, PancakeSwap, Aerodrome, Meteora — skips the scaffolding: you deposit real liquidity, and your seed sets the starting price. The trade-off is symmetrical. The curve gives you a free, standardized, rug-resistant open in exchange for its fixed economics; the pool gives you full control of starting price and depth in exchange for real capital locked in the LP, plus the responsibility of not fumbling it. If you want that comparison priced out per venue, the simulator covers direct pool launches too.
FAQ
What is a bonding curve in simple terms? A vending machine for a new token where every purchase raises the price of the next one, following a formula fixed in the contract. Buys push the price up the curve; sells push it back down the same curve.
Why does the price rise when people buy? The contract prices the token as the ratio of two reserves (quote ÷ tokens). Buying adds to the quote reserve and removes tokens, so the ratio — the price — rises. On constant-product curves the effect compounds: the last 10% of the curve costs multiples of the first 10%.
What are virtual reserves? Fictional starting balances that let the contract price a token as if a pool existed before any real money arrived. Pump.fun starts at 30 virtual SOL vs ~1.07B virtual tokens; those two constants alone determine its entire price path.
What happens when a bonding curve completes? The token graduates: the launchpad uses the raised quote and remaining supply to create a real DEX pool, and trading moves there permanently. On Pump.fun that takes roughly 85 SOL of net buys; see the chart above for every venue's threshold.
Can a bonding curve launch rug? The classic rug — pulling the liquidity — can't happen, because there is no LP to pull and the curve always honors sells. What remains is supply risk: whoever holds a large share of tokens can dump them into the curve. That's a holder-distribution question, not a curve question, and it's why serious teams manage their launch allocation deliberately.
Is a bonding curve the same as an AMM? Mathematically they're cousins — most launchpad curves literally run the AMM formula with virtual reserves. The practical difference: an AMM prices real deposited liquidity forever; a launchpad curve prices a fixed sellable supply once, then hands off to an AMM at graduation.
Sumo runs the whole cycle on every venue in this chart: launching the token (the same wizard math you just played with), trading on the curve before graduation, and the market making, counter trading, and execution that carry it after — with your keys, not a launch dev's.
