Why does Vitalik Buterin prefer PoS over POW?
TechFlow Selected TechFlow Selected
Why does Vitalik Buterin prefer PoS over POW?
PoS systems still offer more benefits than drawbacks: higher efficiency, and stronger capabilities in defending against attacks and recovering from them.
Navigating Web3 tides with focused insightsCompiled by: Chih-Cheng Liang, Hsiao-wei Wang
There are three key reasons why Proof of Stake (PoS) outperforms Proof of Work (PoW) in blockchain security.
Proof of Stake provides higher security at the same cost
The simplest way to understand this is to compare PoS and PoW under the assumption of $1 in daily block rewards—how much would it cost to attack such a network?
GPU-based Proof of Work
You can cheaply rent GPUs, so the cost of attacking the network is merely renting enough GPU power to surpass existing miners. For every $1 in block rewards, current miners' costs approach $1 (if costs exceed $1, miners exit due to unprofitability; otherwise, new miners join to capture profits). Therefore, the cost to attack the network needs only exceed $1 per day—and possibly for just a few hours.
Total attack cost: ~$0.26 (assuming a 6-hour attack), and since attackers receive block rewards, this figure could even drop to zero.
ASIC-based Proof of Work
ASICs represent capital costs: when purchasing an ASIC, you expect it to last about two years before being worn out or replaced by more efficient hardware. If a chain suffers a 51% attack, the community will likely respond by changing the PoW algorithm, rendering your ASIC obsolete. On average, mining costs consist of roughly 1/3 operational costs and 2/3 capital costs (see here).
Thus, for each $1 in block rewards, miners spend ~$0.33 daily on electricity and maintenance, and ~$0.67 on their ASICs. Assuming an ASIC lasts approximately 2 years, miners must invest $486.67 per unit of ASIC hardware. (Note: $486.67 = 365 days × 2 × $0.67 capital cost)
Total attack cost: $486.67 (ASICs) + $0.08 (electricity and maintenance) = $486.75
(Note: Electricity and maintenance costs also assume a 6-hour attack duration.)
That said, it's important to note that ASICs (compared to GPUs) bring higher security at the high cost of centralization, making entry into ASIC mining extremely difficult.
Proof of Stake
The cost of Proof of Stake is almost entirely capital cost (staked coins); the only operational cost is running a node. How much capital would people be willing to lock up for $1 in daily block rewards? Unlike ASICs, staked coins do not depreciate, and when users wish to stop staking, they can withdraw their stake after a short waiting period. Thus, participants should be willing to commit more capital relative to ASIC-based systems for the same level of reward.
Let’s assume a ~15% return is sufficient incentive to stake (this is eth2’s target yield). Therefore, $1 in daily block rewards would attract staking equivalent to 6.667 years of returns, or $2,433 in value. The hardware and electricity costs for nodes are minimal—a $1,000 computer can stake thousands of dollars’ worth of assets, with monthly electricity and internet costs around $100 being sufficient. Conservatively, assume these recurring costs account for ~10% of total staking costs. Hence, only $0.90 of the daily block reward corresponds to capital cost, requiring us to reduce the above number by ~10%.
(Note: 6.667 years = $1 / (15% annual return); $2,433 = $1/day × 365 × 6.667)
Total attack cost: $0.90/day × 6.667 years = $2,189
Long-term, this attack cost is expected to rise further as staking becomes more efficient and people accept lower yields. I personally expect this figure to eventually reach around $10,000.
The only "cost" of achieving this high level of security is reduced liquidity during the staking period. It’s even possible that the perceived scarcity of locked-up coins increases their market value, keeping the total circulating supply or productive investment capital unchanged. In contrast, PoW’s consensus "cost" is the massive waste of electrical energy.
Higher security or lower cost?
Note that we have two ways to use this 5–20x increase in security per dollar spent. One is to maintain current block rewards and benefit from increased security. The other is to maintain current security levels while drastically reducing block rewards (i.e., reducing the "waste" of consensus costs).
Both approaches are viable. Personally, I prefer the latter because, as discussed below, successful attacks cause less damage and recovery is easier under PoS than under PoW.
Proof of Stake enables easier recovery from attacks
In a PoW system, what do you do if your chain suffers a 51% attack? So far, the only practical response has been “wait patiently until the attacker gets bored.” But this ignores a more dangerous threat: the “spawn camping attack,” where an attacker repeatedly targets the chain with the explicit goal of rendering it unusable.
(Note: Spawn camping is a gaming term referring to ambushing an opponent right after they respawn, killing them immediately and preventing any meaningful gameplay.)
GPU-based systems have no defense against this, and persistent attackers can easily render a chain permanently useless (or more realistically, force migration to PoS or proof-of-authority). After the first few days of attack, the attacker’s ongoing costs become very low, while honest miners leave because they cannot earn block rewards under sustained attack.
In ASIC-based systems, communities can respond to the first attack, but subsequent attacks become easy. After the initial attack, the community can hard fork to change the PoW algorithm, effectively bricking all ASICs (both attackers’ and honest miners’). But if attackers are willing to absorb the cost of bricked ASICs, the situation becomes identical to the GPU case (since there isn’t enough time to manufacture new ASICs for the new algorithm), allowing attackers to cheaply conduct repeated spawn-camping attacks afterward.
Note: “Bricking” is slang for electronic devices becoming non-functional, like a brick.
In PoS, the situation is much brighter. For certain types of 51% attacks—specifically those attempting to reverse finalized blocks—PoS includes built-in slashing mechanisms that automatically destroy a large portion of the attacker’s stake (without affecting others’ stakes).
For other, harder-to-detect attacks (e.g., 51% collusion to censor others), the community can coordinate a minority user-activated soft fork (UASF), which can massively destroy attacker funds (in Ethereum, via “inactivity leak”). There’s no need for drastic measures like hard-forking to delete currency. Aside from requiring human coordination to select which minority chain to follow, everything else is automated and rule-compliant.
Note: A minority block is one decided by validators holding less than 51% of the total stake.
Therefore, the first attack on the chain would cost the attacker millions of dollars, and the community could stabilize within days. A second attack would still cost millions, as the attacker must buy new coins to replace previously burned ones. A third attack would burn millions more. The asymmetry is extreme—and advantage lies firmly with the defenders.
Proof of Stake is more decentralized than ASIC-based mining
GPU-based PoW is reasonably decentralized because acquiring GPUs isn't too difficult. However, as previously noted, GPU mining struggles to meet the “security under attack” criterion. In contrast, ASIC mining requires millions in capital—and if you’re buying ASICs, manufacturers often hold disproportionate advantages.
This capital barrier answers the common critique that “PoS favors the rich”: ASIC mining also favors the rich, and in that scenario, wealth concentration is even more pronounced. PoS has a relatively low minimum staking threshold, making participation far more accessible to ordinary individuals.
(Note: At the time of writing, with ETH priced at $440, the minimum staking threshold is approximately 93,000 RMB.)
Furthermore, PoS is more resistant to censorship. GPU and ASIC mining are easy to detect—they require massive power consumption, expensive hardware purchases, and large facilities. In contrast, PoS can run on an inconspicuous laptop, even over a VPN.
Potential advantages of Proof of Work
I believe PoW has two main advantages, though both are significantly limited.
PoS is more of a “closed system,” potentially leading to greater long-term wealth concentration.
In PoS, if you hold coins, you can stake them and earn more of the same coin. In PoW, you always need external resources (electricity, hardware) to acquire more coins. Hence, some argue that PoS leads to more concentrated coin distribution over time.
My response is that PoS rewards are generally low (so validator earnings are low). In eth2, validator yields are expected to be ~0.5–2% of total ETH supply annually. As more validators participate, yields decrease further. Thus, it might take a century for asset concentration to double—and over such timescales, countervailing forces (people spending money, distributing assets to charity or heirs, etc.) are more likely to dominate.
PoS requires “weak subjectivity,” whereas PoW does not.
The concept of “weak subjectivity” is explained in its original introduction. Essentially, when a node first joins or rejoins after being offline for months, it must rely on external sources (friends, exchanges, blockchain explorers, client developers, etc.) to identify the correct chain head. PoW does not have this requirement.
However, this may be a very weak requirement. In practice, users already need to trust client developers or the “community” to this extent. At minimum, users must trust someone (usually client developers) to inform them of the protocol rules and updates. This is unavoidable in any software application. Thus, PoS’s marginal trust cost remains very low.
Even if these risks materialize, PoS systems remain net positive in my view: greater efficiency, stronger attack resilience, and faster recovery capabilities.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@VitalikButerin
