Why Are There No Big DApps on Ethereum?

On July 30, 2025, we will be celebrating a decade since Ethereum launched on mainnet. Inarguably, one of the biggest milestones in this industry’s short life.

When it launched as the world’s first smart contract platform, this was obviously something entirely new and a completely new way of thinking about software. Instead of renting access to someone else’s platform that could change the rules or lock you out at any moment, one could – in theory – now participate in systems that belonged to everyone and no one, where the rules were written in code and couldn’t be arbitrarily changed by a CEO’s whim. Users would own their date, and software would be maintained and managed by a network rather than a boardroom. The consequences seemed pretty utopian.

However, nearly ten years on from Ethereum’s launch and the dreams of a Web3 version of Amazon, eBay, Facebook or TikTok haven’t arrived, and are nowhere on the horizon.

Gavin Wood, Ethereum co-founder, and his vision of envisaged exactly that. Joe Lubin, the renowned founder of Consensys, said that “Ethereum will have that same pervasive influence on our communications and our entire information infrastructure.”

The libertarian journalist Jim Epstein a year after Ethereum’s launch that “the same types of services offered by companies like Facebook, Google, eBay, and Amazon will be provided instead by computers distributed around the globe.”

Vitalik Buterin himself envisaged Ethereum “law, cloud storage, prediction markets, trading decentralized hosting, [hosting] your own currency,” in his 2014 Bitcoin Miami , where he announced Ethereum to the world. “Perhaps even Skynet,” the fictional artificial neural network from the Terminator films. He has described the platform he created as both a to platforms like Facebook and Twitter back in 2021.

The barrier to achieving this vision is scale. The most successful consumer applications today serve hundreds of millions of users. Instagram processes more than 1 billion photo uploads daily. eBay handles roughly 17 billion dollars in transactions each quarter. Facebook’s messaging platforms process trillions of messages annually.

Ethereum processes about 14 transactions per second, and Solana can handle over 1000. Instagram handles over 1 billion photo uploads daily. eBay processes 17 billion dollars in transactions quarterly. The math doesn’t work.

Let’s entertain the decentralized eBay example for a moment. A truly decentralized eBay would demand far more than simple payments. Every listing creation or update would require onchain transactions for item metadata, pricing, and condition details. Auctions would need automatic bidding resolution with time-locked smart contracts. Escrow systems would have to hold funds until delivery confirmation, with DAO arbitration for disputes.

User reputation systems would require immutable rating storage tied to wallet addresses. Inventory management would need real-time stock tracking, possibly through tokenized goods. Shipping confirmations would demand oracle integration for delivery proofs. Marketplace fees and tax royalties would need smart contract enforcement. Optional identity verification systems would require decentralized credential management. Each interaction would multiply the transaction load exponentially beyond what current infrastructure could support.

It goes without saying that this would require a blockchain of unprecedented speed and throughput. Frankly, a decade after Ethereum, the infrastructure just hasn’t been there to support it.

The business model hasn’t always made sense either. Modern applications need massive scale to generate revenue that covers development costs. Furthermore, layer 2 solutions fragment users across platforms, where (for example) Arbitrum users can’t directly interact with Polygon applications. This defeats the purpose of building unified global computing.

This isn’t theoretical. OpenSea despite dominating NFT trading with high-value transactions & fee-tolerant users. If you can’t profit from selling digital art to crypto enthusiasts paying hundreds in fees, how do you build a marketplace for used goods? The economics are even worse for lower-value transactions that define mainstream commerce. A decentralized social network charging $5 per post would be dead on arrival.

Gaming applications that require a few dollars in transaction fees for every item trade won’t attract players who expect the same for free elsewhere. So far, the only viable on-chain businesses have been those that can extract massive value from relatively few users – essentially high-stakes financial applications and speculative trading.

The industry accepted a false tradeoff: security and decentralization, or functionality and scale, but not both. But transaction throughput has steadily increased (and will ) across networks as the technology matures. We can now achieve massive scale even with proof of work chains, maintaining the security and decentralization that made blockchain revolutionary in the first place (rather than the premature embrace of proof of stake that compromised these principles).

Zero-knowledge proofs allow users to prove transaction validity locally, submitting only small cryptographic proofs that are aggregated recursively and in parallel by a network of provers. Networks can process millions of transactions without every node verifying each one individually. When users prove their own transactions, the marginal cost of adding an additional transaction approaches zero, and blockchains can finally support the economics that mainstream applications require.

But ten years on, it’s clear that the vision once laid out by the futurists of Web3 has moved at a disappointing pace. Let’s hope the next decade moves a little faster – and, fingers crossed – our blockchains too.