Reading the room in a room of code—when a single tweet about Conflux’s new Layer-2 scaling upgrade goes viral, I don't get excited. I get suspicious. Last week, a cryptic post from a pseudonymous account claimed that Conflux’s upcoming zkEVM rollup had achieved 90% of Ethereum’s throughput on testnet. No benchmarks. No transaction hash. Just a graph with no axis labels. The post exploded in Chinese crypto circles, even being cited by a minor news aggregator as evidence that “Chinese Ethereum Killers are back.” But as someone who built Python scripts to verify Zcash proofs back in 2020, I know that claims without code are just noise.
I don’t. Not blindly. Let’s dissect.
Context: The Conflux Narrative Conflux has always been a curious beast in the Chinese blockchain scene. Its tree-graph consensus algorithm promised high throughput without compromising security, but its token value and developer activity have lagged behind Ethereum L2s like Arbitrum and Optimism. In 2023, Conflux announced a partnership with China’s state-backed blockchain network, BSN, giving it an air of legitimacy. But the ecosystem remained quiet. Now, in mid-2026, with the modular thesis dominating discourse, Conflux’s narrative shift to a zkEVM Layer-2 is a bid for relevance. The problem? The data doesn’t back it up.
Core: The Data That Wasn’t There Drawing from my experience auditing cross-chain bridges during the 2022 bear market, I know that performance claims are only as strong as their verification layer. The viral post lacked several critical pieces:
- Transaction finality time: The graph showed TPS but omitted the confirmation time. Conflux’s original chain had a median finality of ~30 seconds. If the L2 inherits this, even high TPS is meaningless for DeFi composability.
- Data availability cost: The post claimed the rollup uses Conflux’s native chain for DA. But Conflux charges gas fees in CFX, and at “peak throughput,” the DA cost per transaction would skyrocket—potentially making it more expensive than Ethereum L2s. I’ve seen this pattern before: a bullish narrative hiding a broken unit economics.
- No open-source code: The zkEVM circuits were not published. Without code, the claim is a ghost. I can’t verify the proofs, and neither can the market.
Using Python’s web3 library, I attempted to find any testnet contract on Conflux’s test network. There was none. The only on-chain activity was a single NFT mint from a known Conflux whale wallet. The entire narrative was built on a single image and a coordinated Twitter thread. This is classic behavioral crypto-anthropology: when data is absent, emotions fill the void. The viral post preyed on the “China vs. US” competition narrative, tapping into nationalist pride.
Contrarian: The Real Gap Isn’t Technology—It’s Ecosystem The counter-intuitive angle: even if Conflux’s L2 matched Ethereum’s throughput, it wouldn’t challenge Ethereum or its L2s. Why? Because the bottleneck isn’t TPS—it’s developer mindshare and composable liquidity. Ethereum has a flywheel of developers building applications that share a unified state across L2s via interoperability standards (like CCIP or native bridging). Conflux has no such network effect. Its most popular dApp is a tokenized RPG with 200 daily active users. The obsession with “closing the gap” ignores that the gap is not technical; it’s sociological. I don’t.
Furthermore, the push for “China’s own blockchain” has been tried before—NEO, QTUM, VEChain. Each launched with patriotic fervor and faded into irrelevance because they offered no unique value proposition beyond “We’re Chinese and government-approved.” Conflux’s L2 narrative repeats this pattern. The market’s blind spot is assuming that technical parity equals competitive threat. It does not. Pr
Takeaway: The Next Narrative Is Data Availability, Not Throughput The Conflux episode is a microcosm of a larger market delusion: we fetishize TPS while ignoring the plumbing. The next real narrative shift in crypto won’t be about a single L2 “atching up” to Ethereum—it will be about affordable, secure data availability layers like Celestia and Avail. Conflux, by claiming its own chain as DA, is trying to sell a walled garden. But the market is moving toward modular, permissionless plumbing. I’ve seen this play out in the 2023-2024 DA wars: projects that bet on proprietary DA over shared security lose. So, as I run my script to verify the next claim, I’ll leave you with a question: In a room of code, who is reading the room?