Everyone pitches a faster chain, but nobody asks the harder question: does the dapp need the chain, or the chain need the dapp? Killer apps hold the leverage — and they increasingly build their own chains. Here's the framework I use before believing any new-chain hype.
The relationship between blockchains and the apps that run on them is stranger than most people admit. Ethereum is the oldest ecosystem chain, and its Layer 2 offspring — Arbitrum, Base, Optimism, and the dozens behind them — put on the real fireworks. But the entire reason those L2s exist was one thing: mainnet gas was too expensive. That premise is quietly evaporating.
So before you wire money into the next chain's presale, I want to ask a question almost nobody asks out loud: does the dapp depend on the chain, or does the chain depend on the dapp? The answer decides which tokens survive the next bear market — and which ones are just paying rent to mercenaries.
For years the pitch was simple. Mainnet swaps cost dollars; L2 swaps cost pennies. High-frequency, performance-hungry apps had no choice but to deploy on L2s because they were cheap and fast.
But Ethereum has shipped upgrade after upgrade — Dencun (EIP-4844 blobs, March 2024), Pectra (May 2025), and Fusaka (December 2025) — and mainnet costs have collapsed. One transaction-cost study puts the median mainnet fee at $3.79 in Q1 2024 and just $0.012 by Q1 2026 — a ~99.7% drop. A simple transfer now runs about a cent on a quiet day.
L2s are still cheaper — median L2 fees fell from $0.18 to roughly $0.0015 over the same window — so they hold roughly an 8-10x edge. But notice what happened to the shape of the argument. The old gap was 20x to 100x; today it's a single order of magnitude, and the absolute difference is a fraction of a cent. For a user, "one cent vs. one-tenth of a cent" is not a reason to bridge anywhere. The moat L2s were built on is draining.
Look at what actually makes up a chain's ecosystem. There's a standard starter pack: a DEX, a lending market, a launchpad, a bridge. Every new chain has these on day one. None of it is hard — it's the same open-source code, forked. Cloning Uniswap or Aave is a weekend project, and thanks to rollup-as-a-service vendors like Conduit, Caldera, and AltLayer, spinning up a whole L2 is now a near-one-click product. Caldera alone lets you deploy across five different frameworks.
If the base layer is copy-paste, it can't be the thing that makes a chain matter. So what does?
The thing that makes a chain matter is the killer app — the product people show up for. pump.fun is Solana's. It became the first Solana app to cross $1B in cumulative revenue (March 2026), and in Q1 2026 alone it generated ~$124.7M — over a third of all Solana app revenue. Polymarket is Polygon's — the prediction market that defined an entire category.
Here's the uncomfortable part for chains: these apps bring the users, so the apps hold the power. When the chain underperforms, the app leaves. Polygon PoS suffered a genuine ~50-minute block-production halt in July 2025 (a Heimdall consensus bug), followed by more instability through the autumn. By December 2025, Polymarket's team was openly planning to migrate off Polygon to its own Ethereum L2, citing exactly those infrastructure problems — they need more block space, cheaper gas, and smaller block times than a general-purpose chain gives them.
And increasingly, the biggest apps skip the "rent, then leave" cycle entirely and build their own chain from the start:
| App-specific chain | Focus | Launched | Why build their own |
|---|---|---|---|
| HyperEVM (Hyperliquid) | Trading / DeFi | Feb 2025 | Order-book latency and throughput a shared chain can't match |
| Plasma (XPL) | Stablecoin transfers | Sep 2025 | Zero-fee USDT payments; backed by Tether + Founders Fund |
| Arc (Circle) | Payments / cross-border | Aug 2025 | Sub-second finality tuned for stablecoin settlement |
| Robinhood Chain | Tokenized stocks | Jul 2026 | 24/7 equity trading on its own Arbitrum-stack L2 |
Notice the pattern: when a product knows it will bring serious traffic, it doesn't negotiate for block space — it owns the block space. The app is the scarce asset; the chain is the commodity.
Now consider the ordinary DeFi protocol — a yield vault, a staking wrapper, a mid-size lending market. These have low interaction frequency. A user deposits, maybe rebalances monthly, and otherwise sits still. A few thousand to tens of thousands of users, no extreme performance demands. Mainnet or an L2 — either can carry that load. For this kind of dapp, a chain's honest position is: nice if you come, no big deal if you don't.
Except in a bear market, these mid-tier protocols become the belles of the ball. Chains dip into their own treasuries and dangle token incentives to lure them over — bridge some liquidity to us, the yield is even a little higher here. This cycle it's been everywhere: yield-bearing dollars like USDe, USDai, USDat, and APYX have each bridged a slice of their supply to newer chains such as Monad and BNB Chain, where the advertised yield runs a notch above mainnet. That extra spread doesn't come from real activity — it's paid out of the destination chain's foundation subsidy. Ethena's USDe is the clearest documented case: when it expanded to BNB Chain, liquidity providers were offered boosted "30x Ethena rewards" stacked with CAKE emissions on PancakeSwap, well above what the same capital earned on mainnet. USDe now spans Ethereum, BNB, Arbitrum, Solana and more via LayerZero, with supply around $3.9B.
Bridge over for the boosted yield, bridge back when it ends — the subsidy rents liquidity, it doesn't buy loyalty.
But strip away the branding and this is just mercenary capital with extra steps. The surplus yield is paid by the chain's foundation, not by real economic activity. When the subsidy ends — and it always ends — people bridge their assets straight back to mainnet to chase the next reward. Why? Because the new chain has no native liquidity, and protocols don't want their liquidity fragmented across ten chains anyway. You pay a little gas and come home. The chain spent real money and rented users who were never going to stay.
New chains love to lead with numbers: how many transactions per second, how much parallelism, how many concurrent users. Monad launched mainnet in November 2025 claiming 10,000 TPS (with benchmarks around 5,000+ under load). MegaETH went live in February 2026 claiming 100,000+ TPS — though its public stress tests demonstrated closer to ~35,000, a reminder that claimed and shown are different animals.
The throughput is real engineering. But watch what happens after launch. Mercenary capital floods in to farm the airdrop, on-chain metrics look spectacular for a few weeks, and then the airdrop ships. The farmers leave, activity flatlines, fee revenue evaporates, and the chain goes quiet. High TPS with no durable demand is a stadium with great acoustics and no team playing in it.
The irony is that when real traffic finally shows up on a new chain, it's rarely the traffic the founders pitched. Robinhood Chain launched in July 2026 selling a serious thesis — 24/7 tokenized US equities. Its first genuine breakout was a memecoin: CASHCAT, named after the discarded startup name Robinhood's founders used before they were Robinhood. It ripped roughly 1,700% in 24 hours to a ~$100M market cap, one trader famously turned $800 into over $1M, and CEO Vlad Tenev — previously cool on memecoins — did a public about-face within days, conceding the chain "works great for meme tokens." A chain built for stocks got its first heartbeat from a cat coin. That's not an accident; it's the pattern.
The wealth-effect flywheel: visible fortunes are the marketing, and the crowd keeps the wheel spinning — until the wealth stops.
Solana is the counter-example — the chain that turned that heartbeat into a pulse. Its sustained activity rode a memecoin wealth effect: in early 2025, memecoins were ~44% of Solana's DEX volume and the vast majority of speculative trading. And this wasn't a passive accident — the Solana ecosystem leaned into it deliberately, treating the memecoin casino as a user-acquisition funnel. Cheap, instant launches, a launchpad flywheel, and a culture that celebrated the wins pushed prices higher and pulled in ever more players; the visible fortunes were the marketing. People made money, told their friends, and stuck around; richer infrastructure and more builders followed, and the users slowly took root. The flywheel is powerful — but fragile. When the memecoin economy cracked in February 2026, weekly Solana DEX volume fell ~62% in three weeks (from ~$118B to ~$44.5B). A wealth effect is a real moat right up until the wealth stops.
The clearest way I've found to think about all this is an analogy to cities, companies, and workers.
Imagine New York gets too crowded, so satellite cities spring up to absorb the overflow. For a satellite city to grow, the mayor has to go out and recruit — tax breaks, cheap land, subsidies. A few companies set up shop and start hiring. As workers arrive, a service economy grows around them: restaurants, schools, hospitals, all serving those employees. The ecosystem thickens, more people come looking for work, more companies come to do business, and the whole city takes off. It's a flywheel — but it needs a push, and it needs time.
| The metaphor | In crypto |
|---|---|
| City / satellite town | Blockchain (L1 / L2) |
| Mayor recruiting with tax breaks | Chain foundation subsidizing TVL |
| Company that relocates for lower taxes | A dapp bridging for higher yield |
| Star company that builds its own town | Killer app launching its own chain (HyperEVM) |
| Restaurants, schools, hospitals | DEX, lending, bridges — the service layer |
| Workers who put down roots | Users who stay after incentives end |
Some star companies — think an electric-car maker fleeing a high-tax state — relocate and bring their own gravity: jobs and tax revenue arrive on day one. Some celebrity mayors draw crowds by reputation alone. And some companies, knowing they'll generate enormous demand, skip the search and found their own city — which is exactly what HyperEVM and Plasma did. Meanwhile, as big companies leave, New York's mayor keeps expanding capacity, so more firms just stay put and deepen their roots — friends and family are already there, everything's a half-hour drive. That's Ethereum mainnet raising its gas limit while its L2s compete for tenants.
The lesson: mercenary capital is a double-edged sword. A chain can absolutely use subsidies to bootstrap, but the subsidy has to last long enough for the native ecosystem and infrastructure to actually grow up around it. Only then can the rewards taper off without the users leaving with them. Cut it too early and you're left with an empty stadium.
Which brings me to the structural problem: the supply and demand are badly mismatched. There are too many chains and too few apps. Everyone is subsidizing to attract users. The apps eat well, the users eat well, and the chains do all the work and end up with almost no real, sticky users to show for it.
That's the state of the industry right now. There simply aren't enough dapps to go around, while L2s are a dime a dozen — cloning a rollup is as turnkey as forking Uniswap. A breakout product that realizes it will get big and needs more throughput will build its own chain, like Hyperliquid did. And the small-to-mid dapps mostly camp on Ethereum mainnet, bridging a slice to whichever chain is paying this week. If mainnet performance keeps improving, mainnet becomes a siphon on every L2. Purpose-built apps go off and run their own chains; the generic L2s and their copy-paste ecosystems wither.
That's my read on the dapp-versus-chain relationship. When you're evaluating a new chain, watch three things:
So the next time someone tells you how much they raised, which technical legends are on the team, which big company backs them, which crypto luminary endorsed them, how many hundreds of thousands of TPS the testnet hit — and by the way, the presale is open, send money now — pause and run those three questions first. You may reach a very different answer.
This article is for research and educational purposes only and is not financial advice. Do your own research before interacting with any protocol or participating in any token sale.
It depends on the dapp's leverage. Commodity DeFi protocols (yield vaults, mid-size lending) need whichever chain pays them, so chains court them with subsidies. But killer apps like pump.fun or Polymarket bring the users, so the chain needs them — and when a chain underperforms, these apps leave or build their own. The scarce asset is the app; the base-layer chain is largely a copy-paste commodity.
No. After the Dencun, Pectra and Fusaka upgrades, the median mainnet transaction fee fell from about $3.79 in Q1 2024 to roughly $0.012 by Q1 2026 — a ~99.7% drop. A simple transfer now costs about a cent on a quiet day. L2s are still cheaper (median ~$0.0015), keeping an ~8-10x edge, but the absolute gap has shrunk to a fraction of a cent — weakening the original reason L2s existed.
Because they generate enough traffic that a shared chain's latency, throughput and reliability become a bottleneck. Hyperliquid launched HyperEVM (Feb 2025) for order-book performance; Polymarket announced plans to leave Polygon for its own L2 after a ~50-minute Polygon halt in July 2025. When an app knows it will bring serious demand, owning the block space beats renting it.
Mercenary capital is yield-chasing money that parks wherever incentives are highest and leaves the moment they dry up. New chains lure protocols with foundation-funded subsidies — like the boosted rewards Ethena's USDe got on BNB Chain — but that surplus yield isn't from real activity. When the subsidy ends, capital bridges back to mainnet, and the chain is left having paid real money for users who never intended to stay.
Not on its own. Chains like Monad (claims 10,000 TPS) and MegaETH (claims 100,000+, demonstrated ~35,000) have real engineering, but throughput without durable demand is an empty stadium. New chains often see spectacular metrics for weeks as mercenary capital farms an airdrop, then activity flatlines once rewards ship. Sustained success comes from sticky, real-user demand — the way Solana's memecoin wealth effect kept it active — not from a benchmark.
Be skeptical of pitches built on funding raised, team pedigree, big-name backers, or testnet TPS. Before committing, ask three questions: Will more sticky dapps actually appear, or are chains multiplying faster than apps? Will Ethereum mainnet keep getting cheaper (siphoning L2s)? And will more star apps build their own chains, draining killer apps from shared chains? If the answers are unfavorable, the chain may struggle to retain real users.
Practitioner turned analyst tracking how incentives, liquidity, and capital flows shape DeFi protocols.
