The road to 10m tx/sec⚡️

How can we trust a settlement of some value?

By verifying the transactions!

To achieve this we have historically invested a huge amount in “social capital”

Banks
Regulators
Accountants
Lawyers

… all of whom stand on the shoulders of giants

Until 1968 Wall Street was drowning in paper from a surge in trading, which caused the exchange to close often on Wednesdays to settle trades.

0.05 tx/sec

Later trades would be recorded and settling electronically solving most of the scaling problems

100,000-1,000,000 tx/sec

… but settlement latency is often 3 days

Since 2009 another way to achieve the same result has been to perform verifiable computation

10 tx/sec

Chain	Nodes
Ethereum 2.0	238,099
Cardano	2,076
Avalanche	1,023
Tezos	377
Polkadot	297
Cosmos Hub	150
Nano	107

T✌️ o ways this evolves

Interchain

Multiple L1s with bridges
Current bridges are a security risk
Better design will translate consensus data
IBC interchain accounts

Multichain

Rollups
Shared security

… or more likely a linear combination of those two

Specialization reverses the blockchain trilemma

Consensus 🤝

Execution 🧠

Data availability 💾

Consensus 🤝

Ordering of transactions
Executing settlement logic
Decentralization matters

100 tx/sec

Execution 🧠

Single operator without compromising on security
Optimistic rollups with fraud proofs
Zero-knowledge rollups with validity proofs

10,000 tx/sec

100 * 10,000 = 1,000,000 tx/sec

Data availability 💾

Data sharding
Initially 16 shards
Scaling up to 1024 shards

30% cost reduction

Rollups (L2)

Merkle tree

Zero-knowledge Rollups

Cross-rollup communication
Shared state, including liquidity
Security guarantees
zkEVM

zkEVM on L1

Running on mobile phones and in browsers

70% cost reduction

Now we just need an unforeseen 10x in throughput and then we’re up to 10 million transactions per second with a settlement latency of a few seconds for next to nothing 🤔