The P/S that nobody applies to blockchains: fees vs. value

Bitcoin is trading at 13,529 times the fees its network generated in the twelve months leading up to July 2026; Hyperliquid's perpetuals engine, at 14. Price-to-sales — the ratio that divides capitalization by annual revenue, the first metric a stock analyst looks at before profit — is routine for valuing a stock, and platforms like Token Terminal or Delphi Digital already publish their version applied to crypto. The differentiator of this article lies in two actions that almost no one combines: dating the quotient and reading it through two windows simultaneously. We calculate it for seven chains with explicit arithmetic across two timeframes (last 30 days annualized and last twelve months) and use the distance between them as a signal of acceleration or cooling. Additionally, we break down the Hyperliquid case, where separating the base layer gas from trading fees moves the multiple from 1,254 to 14. The result ranks which capitalizations are discounting decades of future growth and which are already trading like a mature business.

What is Price-to-Sales and why does it make sense to apply it to a chain?

In stock analysis, the price-to-sales ratio divides a company's market capitalization by its annual revenue to produce a multiple. Applied to a chain, the ratio is read the same way: a P/S of 44 —Arbitrum's on July 20, 2026— means the market pays 44 dollars for every dollar that the network generates annually in fees; Bitcoin's 13.529 on that same day means it pays 13.529. Aswath Damodaran, the New York University valuation professor whose tables are used by half of Wall Street, describes it as the most difficult multiple to manipulate because sales are the top line and the hardest to window-dress on an income statement. We review this and other indicators in our guia de métricas DeFi.

A blockchain has its own sales line: the fees users pay to use it. Every USDT transfer on Tron, every swap on Solana, every data block on Arbitrum leaves a fee measured in the native token. DefiLlama aggregates this in dollars and publishes it daily. The market cap of the native token is the equivalent of the market value of shares. Dividing one by the other gives a comparable multiple across chains that are otherwise incomparable architectures: there is no way to directly compare Bitcoin's consensus with an L2 on Ethereum, but you can ask how much the market pays for every dollar of fees each one produces.

How is a blockchain's P/S calculated step-by-step?

The formula is a simple division, and both inputs are just a couple of API calls away. The steps:

• Chain Fees. In api.llama.fi/overview/fees, filter by the "Chain" category: these are the native gas fees paid by those using the network, not the fees from protocols running on top of it. DefiLlama returns the 30-day aggregate (total30d) and twelve-month aggregate (total1y).
• Annualize the short window. Multiply the 30-day fees by 365/30 (≈12.17) to bring them to an annual pace. This allows for an apples-to-apples comparison against the twelve-month series.
• Circulating Market Cap. In api.coingecko.com/api/v3/coins/markets, take the market_cap field of the native token. This is the circulating supply, not the fully diluted valuation (FDV), and that choice matters (more on this below).
• Divide. P/S = market cap ÷ annualized fees. Do this for both windows and you will have two multiples: one reflecting the current momentum and another for the entire year.

Example with Solana on July 20: 30-day gas fees of 14.32 million dollars, which annualized result in 174.2 million; twelve-month fees of 283.5 million; circulating capitalization of 44,347 million. The 30-day annualized P/S is 44,347 ÷ 174.2 = 255; the twelve-month P/S is 44,347 ÷ 283.5 = 156. Two numbers for the same chain on the same day: the 99-point gap between them measures how much Solana gas has cooled over the last four weeks compared to its annual average.

What does the July 20, 2026 table say?

Seven chains with native tokens, ordered from the cheapest multiple to the most expensive based on the twelve-month window. All use native gas fees from DefiLlama's "Chain" category so the methodology remains identical for every row. Hyperliquid appears here with its base layer gas; its special case is broken down further below, as that is where the metric cracks.

The range spans from 44 to 13,529 over the twelve-month window: two and a half orders of magnitude between the chain the market prices cheapest per dollar of fees and the one it prices most expensive. The ranking does not resemble the price or market cap rankings. Tron, the public chain generating the most gas fees on this list (409.6 million per year, more than Ethereum), trades at a multiple of 75, while Ethereum — which generated less native gas in the twelve months leading up to July 2026 — trades at 847, eleven times higher. Ethereum trades at 847 compared to Solana's 156: more than five times more expensive per dollar of fees, despite Solana generating more gas over the annual window. Base enters as a footnote: its network generated 62.0 million in gas fees over twelve months, but since it lacks a native token, it does not support a P/S ratio and remains excluded from the main table despite generating revenue like a mid-sized chain.

Why do Bitcoin and Ethereum trade at four and three-figure P/S ratios?

Because the market is not buying its fees. It is buying something else, and the multiple gives it away. Bitcoin at 13,529 times its annual fees is absurd if you read it as a transaction processing business: nobody pays thirteen thousand years of sales for a company. But Bitcoin's market cap does not discount fees; it discounts its role as a reserve asset —digital gold, institutional collateral, treasury reserve— where network fees are merely a footnote to the investment case. That extraordinary multiple serves as numerical proof: it quantifies the extent to which Bitcoin's valuation relies on its role as a reserve and hardly at all on the 95.2 million in fees that its chain generated in the twelve months leading up to July 2026.

Ethereum, at 847, occupies an awkward middle ground. Its capitalization discounts that it will become the settlement layer for a tokenized economy much larger than the current one: real-world assets, stablecoins, and L2s that return value to the base layer. It is a growth narrative spanning decades, and a triple-digit P/S is what those who believe that story pay in advance. Solana (156) and BNB (353) discount more modest versions of the same tale. At the other extreme, Tron (75) and Arbitrum (44) trade almost like mature businesses: the market pays for the fee flow that already exists, not for one ten times larger arriving in 2035. That Arbitrum appears "cheap" does not mean it is a good investment — it means the market attributes no future growth to it and has punished its token down to a capitalization of 572 million. P/S measures expectations baked into the price, not quality.

Why don't the 30-day and 12-month windows match?

In the table, almost all chains have a 30-day P/S higher than their twelve-month P/S. This means that their fees over the last four weeks are running below the pace of the full year: 2026 is proving to be a lower-fee environment than the previous fiscal average. On Ethereum, the gap is massive —2,441 versus 847— a sign that its recent native gas has plummeted compared to last year, consistent with a cheap block space market following the expansion of batch data. In Solana, the picture repeats (255 versus 156) and in Arbitrum, it becomes even more pronounced (109 versus 44).

The direction of the divergence is information in itself. A 30-day P/S significantly above the twelve-month figure warns that the recent picture is worse than the annual average, and that relying solely on the trailing figure overestimates current activity. The inverse reading —recent fees above the annual average— does not appear today in any of the seven: as of July 20, even the chain that breaks the table in the next chapter is running below its annual average, whereas two weeks earlier it was the only one accelerating.

Why does Hyperliquid break the table?

Because on Hyperliquid, base layer gas is irrelevant and the entire economy lives one layer above. Its L1 collected barely 10.8 million in gas fees over twelve months, placing it at a P/S of 1,254: it would appear to be the second most expensive chain on the list. But Hyperliquid is not a general-purpose network that charges for generic transactions; it is a perpetuals exchange (futures without an expiration date) built on its own chain. The business is trading fees, and DefiLlama accounts for them separately: 987.0 million dollars over twelve months, with the last four weeks running at an annualized rate of 668.7 million —below the yearly average: the 2026 cooldown is also reaching the leader, which in early July was still the only chain in the sample accelerating—.

With that figure, Hyperliquid's P/S drops from 1,254 to 13.7 on a twelve-month window and to 20.2 on a 30-day annualized basis: moving from one of the most expensive on the table to the cheapest by a factor of more than ninety. The lesson is methodological: the multiple depends entirely on which fee line you count, and that nearly hundred-fold jump proves it within a single chain. For a monolithic L1 like Solana or Tron, native gas is the business. For Hyperliquid, gas is the trivial toll for a parking lot where the real revenue is in the shop upstairs. Comparing Ethereum's gas P/S with Hyperliquid's gas P/S is comparing two things that share a name but measure different businesses. We cover that perpetuals engine in detail in our analysis of Hyperliquid revenue fundamentals.

Are a chain's fees income for the holder?

This is the serious limit of the analogy, and it shouldn't be hidden. When a company invoices, that money belongs to the corporation and, via profit, to its shareholders. When a chain collects fees, the money goes to the validators or miners producing the blocks, not automatically to the token holder. A stock's P/S relies on the fact that sales eventually end up, in part, in the owner's pocket. In a chain, that only happens if there is a mechanism that redirects the fee toward the holder: burning or distribution.

And there, the seven chains diverge radically:

• Ethereum burns the base fee of every transaction (since the 2021 EIP-1559 update), removing it from circulation: fees do reduce supply and indirectly benefit the holder.
• Solana burns 50% of every fee and gives the rest to the validator.
• BNB destroys tokens quarterly and automatically, tying supply to activity.
• Hyperliquid allocates a significant portion of its exchange fees to buy back its own token on the open market—a buyback—and back it via its insurance fund, the most direct link between invoicing and holder value on the entire list.
• Bitcoin gives the entire fee to the miner: zero reaches the holder through this route, reinforcing that its valuation does not rely on fees.

That is why the same P/S means different things depending on the chain: 154 on Solana, with partial burning, is not equivalent to 154 on a chain that burns nothing. The ratio measures how much the market pays for the fee flow; how much of that flow actually reaches the holder is a second question—that of the value capture mechanism: burning (permanent removal of tokens from circulation), buybacks, or a fee-switch (a fee toggle that redirects part of the invoicing to the token). Without that second data point, P/S is a necessary but not sufficient condition to talk about value for the token holder. We develop the distinction between gross fees and revenue retained by the protocol in our DeFi real revenue ranking.

What does a chain's P/S really measure and what does it not?

It measures narrative paid for in advance. A P/S of 44 indicates that the market is paying for the existing fee flow; one of 13,529 suggests that fees are irrelevant to the thesis and that something else is being bought —store of value, regulatory optionality, future dominance—. Between these extremes, each multiple quantifies how much growth is priced in: Ethereum at 847 must multiply its fees many times over to justify its market cap; Tron at 75, much less so.

What the P/S does not do is tell you which chain is better: Arbitrum at 44 could be an undervalued chain or one that the market has left for dead, and the multiple alone does not distinguish between the two readings. That is why the table serves as a diagnostic tool and not a recommendation: use it to see what narrative you are buying when you buy a token, contrast the 30-day window with the twelve-month one to know if that narrative is playing out or cooling off right now, and check if the chain has a mechanism that converts fees into value for you. Fees, market cap, and value capture are public data; what this exercise provides is dating them on the same day and reading them across two windows simultaneously, which is where the signal of acceleration or cooling appears. You can find the market share counterpoint in our analysis of Ethereum dominance by TVL, which measures deposited capital, not fees collected.

Sources and links: DefiLlama — fees by chain · CoinGecko — market caps · Ultra Sound Money — ETH burn (EIP-1559) · Solana Docs — fees and burning · BNB Chain — auto-burn · Hyperliquid Docs — insurance fund · Damodaran — valuation multiples