Ten years of code data, ten valuation criteria, one composite score, three scenarios. No narrative. No price target. Just the formula.
At PFVS 86.5 with the current trajectory (Glamsterdam on devnet, Pectra + Fusaka shipped, stablecoin growth continuing), the realistic path is the most-defensible read of the data. That implies roughly +0% from May 2026 by 2030 — if the market repriced to fundamentals. Confidence is on direction, far less on timing or magnitude.
From Frontier (2015) through Pectra + Fusaka (2025). PoW issuance, Merge transition, EIP-1559 burn, validator entry — every data point that explains why the supply curve looks the way it does.
Each criterion scored 0–100 against a defined ladder. Weights sum to 100%. The aggregate is the Protocol Fundamental Value Score — a relative measure of structural strength.
| # | Criterion | Weight | Score | Contribution |
|---|---|---|---|---|
| 1 | Network Security (NSS) Staked value × cost-of-attack. ETH at 26.5% staked, $45B cost-of-attack. | 12% | 78 | 9.36 |
| 2 | Standards Lock-in (SLS) # canonical standards × years × universality. ERC-20/721/1155/4337 universal across every EVM chain. | 12% | 96 | 11.52 |
| 3 | Multi-chain Anchor (MAS) 100+ L2/L3 chains anchor to L1 for settlement and data availability. | 10% | 95 | 9.50 |
| 4 | Client Diversity (CD) 5 EL + 5 CL independent clients, no >50% dominance. | 8% | 90 | 7.20 |
| 5 | Roadmap Execution (RER) 22 forks, ~150 EIPs activated. The Merge, EIP-4844, EIP-7702, PeerDAS all shipped as committed. | 10% | 88 | 8.80 |
| 6 | Developer Activity (DA) 10 actively-maintained clients, 70+ PROPOSED EIPs in flight, Glamsterdam PFI = 41 EIPs. | 8% | 92 | 7.36 |
| 7 | Tokenized Value Anchor (TVA) $153B stablecoins on L1 (47% global share) + BUIDL, BENJI, OUSG tier-1 RWAs. | 15% | 90 | 13.50 |
| 8 | Protocol Revenue (PR) Annualised EIP-1559 burn ≈ $1–3B. Mid-tier infrastructure SaaS-equivalent revenue. | 10% | 70 | 7.00 |
| 9 | Decentralization (DEC) ~1M validators, 32 ETH permissionless threshold; Lido ~28% concentration is main weakness. | 10% | 75 | 7.50 |
| 10 | Time-to-Replicate Moat (TTR) 10y standards + 5y multi-client + 5y L2 ecosystem + 5y stake = ~10y compound moat. | 5% | 95 | 4.75 |
| PFVS · Protocol Fundamental Value Score | 100% | 86.5 | 86.5 | |
Weighted-average composite. Each criterion's score multiplied by its weight; summed. No magic, no calibration tricks.
The score is a relative measure. To translate into market-cap terms we triangulate across four independent methods (sum-of-parts, P/Revenue, terminal digital-asset capture, PFVS-anchored). The result is the $280B–$865B fair value range shown above — a range, not a number.
Each criterion maps to an existing piece of real-world infrastructure that has already proven it can compound value over decades. Ethereum is not unprecedented — it is the same pattern in code.
ERC-20/721/1155/4337/4626 are TCP/IP-equivalent — open, immutable interfaces. TCP/IP did not get re-monetised but routed the entire economic value of "the internet" through itself. Ethereum's standards do the same for tokenised value.
L1 at ~15 tps is deliberately the SWIFT of digital value: low throughput, high finality, neutral, open. BUIDL, BENJI, OUSG, USDC, USDT — all settle here.
100+ L2/L3 chains use Ethereum's DA + settlement as their substrate — the way SaaS companies use AWS. Per-block blob fees flow back to L1: the AWS bill of the EVM economy, ≈ $1–3B annualised.
Visa charges a percentage of every transaction; the network effect is that every merchant and cardholder is already on it. Ethereum's gas burn is structurally similar — a fee per usage with no way around it if you want EVM-standards composability.
$67B of staked ETH is the security floor. Cost to attack ≈ $45B. Comparable to the inertia of gold's physicality, but measurable in code rather than mass.
5 independent execution clients + 5 consensus clients, all open-source, all maintained by different teams across different jurisdictions. Linux-like resilience. No single team can halt the chain.
Both halves come from the same data. The "is not" list closes off claims the data does not support — protecting the framework from drift into speculation.