The Law of Crypto Scaling — Where is the Hard Cap for DeFi?

Source: Zuoye Waiboshan
DeepSeek R2 did not come out in May as rumored, but was updated to R1 on May 28. Musk's Grok 3.5 also frequently missed tickets, and it is not as good as Starship to hear the real sound.

With the enthusiastic boost of huge amounts of capital, the scaling law in the field of large models has completed its life cycle faster than Moore's Law of chips.

If software, hardware, and even human life span, cities, and countries have their upper limits on scale effects, then the blockchain field must also have its own laws. At the moment when SVM L2 enters the coin issuance cycle and Ethereum returns to the L1 battlefield, I try to imitate the scale law and give an encrypted version.

Ethereum soft scale, Solana hard cap

We start with the full node data scale.

A full node represents a complete "backup" of the public chain. Owning BTC/ETH/SOL does not mean that we own the corresponding blockchain. Only when we download the full node data and participate in the block generation process can we say "I own the Bitcoin ledger". Correspondingly, Bitcoin has also added a decentralized node.

Solana's 1,500 node scale is struggling to maintain a balance between decentralization and consensus efficiency. Correspondingly, the 400T full node data scale leads a number of public chains/L2.

If not compared with Bitcoin, Ethereum is already very good at controlling data volume. Since July 30, 2015, Ethereum has been in a state of rapid development. Since the creation of the Genesis Block, the data volume of Ethereum's full nodes is only about 13 TB, far less than its "killer" Solana's 400 TB, while Bitcoin's 643.2 GB is a work of art.

In the initial design, Satoshi Nakamoto strictly considered the growth curve of Moore's Law, so that the data growth of Bitcoin was strictly limited to the expansion curve of hardware. It must be said that those who later supported Bitcoin's large blocks could not stand, because Moore's Law has entered the edge of marginal effect.

In the CPU field, Intel's 14 nm++ can be called an heirloom. In the GPU field, Nvidia's 50 series has not "significantly exceeded" the 40 series. However, the progress in the storage field, in Yangtze Memory's Xtacking Under the architecture, the scale of 3D NAND stacking has gradually peaked, and Samsung's 400 layers are the expected high point of engineering at present.

In a word, the law of scale will prevent the underlying hardware of the public chain from making great progress. It can even be said that this is not a short-term technical limitation, but will maintain the status quo for a long time.

Faced with difficulties, Ethereum is obsessed with ecological optimization and reconstruction. Trillions of RWA assets are its must-win. Whether it is to follow Sony's self-built L2 or to speed up and embrace the Risc-V architecture, it is not "finding a more extreme software and hardware collaboration", but sticking to its own advantages.

Solana chooses to move towards the extreme speed of light. In addition to the current Firedancer and AlpenGlow, the super-large node scale has actually excluded individual participants. 13 TB hard drives can still be accumulated, and 400 TB is already a pipe dream. 600 GB of Bitcoin can be theoretically satisfied even if Samsung, LG and Hynix factories are fighting fires every day.

The only question is, where is the lower and upper limit of the chain scale?

The limit of the token economic system

AI did not embrace Crypto as expected, which did not prevent Virtuals from advancing in price. Even blockchain on the left hand and AI on the right hand have become fellow travelers of the current US government's MAGA. 5G and the metaverse are both old, and the heroes are still Sun Ge and stablecoins.

Let's briefly discuss the various limit indicators of the token economic system. Bitcoin, under the premise of lack of practical use, has a market value of 2 trillion US dollars, Ethereum 300 billion US dollars, and Solana 80 billion US dollars. We take Ethereum as the standard value, and the limit of the public chain economic system is 300 billion US dollars.

This does not mean that Bitcoin is overvalued, nor that new public chains cannot exceed this value, but it is highly likely that the market performance of a public chain is the current optimal solution, that is, "we believe that the current market performance is the most reasonable existence", so directly selecting this value is more effective than complex calculations. If it is not necessary, do not increase the entity.

We introduce two concepts from the book "Scale":

1. 1. Superlinear scaling: when the scale of a system expands, its output or benefit does not increase proportionally, but grows at a faster rate.
2. 2. Sublinear scaling: when the scale of a system expands, the growth rate of some of its indicators (such as cost, resource consumption, maintenance requirements, etc.) is lower than the linear proportion.

It is not complicated to understand the two. For example, the growth of Ethereum from $1 (2015) to $200 (2017) is super linear scaling, which is much faster than its growth from $200 to $200. ATH (2021) takes about half as long, and the latter belongs to the classic sublinear scaling.

Everything has its limits, otherwise blue whales, elephants, and North American redwoods will surpass themselves, but the earth's gravity is a hard cap that is difficult to cross.

Continue to drill, has DeFi reached its limit?

The scale limit of DeFi can be contained by Ethereum, and the yield rate is examined instead, which is also the core proposition of DeFi. The driving force of entropy increase lies in the extreme pursuit of yield. We give three standards, UST's 20% APY, DAI's 150% over-staking ratio, and Ethena's sUSDe's 90D MA APY 5.51% calculation at this stage.

We can assume that DeFi's yield capture ability has dropped from 1.5 times to 5%. Even if calculated at 20% of UST, DeFi has reached its upper limit.

It should be noted that the trillions of RWA assets on the chain will only reduce the average yield of DeFi, but will not increase it. This is in line with the sublinear scaling law. The extreme expansion of the system scale will not bring about the extreme increase in capital efficiency.

Also note that there is a market motivation for DAI's 150% over-collateralization ratio: I can make extra profits outside the 150% collateralization ratio, so assuming it as a market baseline is my personal opinion, which is not necessarily correct.

We can be rough. At present, the on-chain economic system, with the token economy as the benchmark model, has an actual scale cap of US$300 billion and a yield of about 5%. Again, this is not to say that the total market value or the upper or lower limit of a single token, but the overall scale of tradable transactions is so large.

In fact, you can't sell 2 trillion bitcoins, and even U.S. bonds can't handle such a large sell-off.

Conclusion

Looking at the history of blockchain development since Bitcoin, the discrete trend between public chains has not been bridged. Bitcoin is increasingly decoupled from the on-chain ecosystem, and the failure of the on-chain reputation system and identity system has led to the over-collateralization model becoming the mainstream.

Whether it is stablecoins or RWA, they are all leveraged on-chain assets, that is, off-chain assets naturally have higher credibility. Under the current on-chain scale law, we may also reach the upper limit of scaling law or Moore's Law. It has only been 5 years since DeFi Summer and 10 years since the birth of Ethereum.