Scorecard 2026: The 10 Largest Blockchains Across 5 Fundamentals

Ethereum burned 990.6 ETH in the last thirty days —approximately $1.83 million— while BNB Chain removed $932 million from circulation in a single event on July 15, 2026. Both chains appear together in any TVL (total value locked, the capital deposited in a network's contracts) ranking, yet that ranking says nothing about the aforementioned difference. This article builds a scorecard of the top ten blockchains by deposited capital, crossing five dimensions that no dashboard publishes together: daily active wallets, TVL, token return mechanism (burn or buyback), marginal utility of the native token, and concentration of decision-making power. Each cell includes its source and date, including the uncomfortable cells —Bitcoin with no on-chain governance, Base with no token, Arbitrum with a token that only votes— because they are part of the lesson. By the end, you will have a reusable method to score any ecosystem, including those not on this list.

What does a fundamentals scorecard measure that a TVL ranking doesn't?

The TVL ranking answers a single question: how much capital is parked. It doesn't say how many people use the network, if the native token receives anything in exchange for that activity, what the point of holding it is once you stop speculating, or who can change the rules. A fundamentals scorecard crosses these five series and lets the contradictions emerge: chains with high capital and few people, chains with many people and no token return, chains with generous returns and three validators calling the shots.

It is worth noting the division of labor with our price-to-sales analysis applied to blockchains, published on July 20. That article measures valuation: it divides market cap by annual fees and answers whether the market is paying a high or low price for every dollar of revenue. This one measures usage and governance: how many people, how much capital, how much returns to the token, and who is in charge. These are independent axes —a chain can appear cheap in the first and fragile in the second— so this scorecard does not reproduce the P/S column or market caps. If you are looking for the multiple, it is there; if you are looking for usage fundamentals, stay here.

Which ten chains make the cut and why is Provenance excluded?

The selection criteria are part of the method and are declared before looking at the results:

• Top-10 by TVL according to the DefiLlama /v2/chains endpoint, captured on August 3, 2026.
• Cap of three Ethereum Layer 2 networks (L2: chains that execute transactions off Ethereum and settle on it to reduce costs). Without this cap, an L2 bull cycle would turn the ranking into an intra-Ethereum list and stop comparing architectures.
• Reasoned exclusion of Provenance ($1.557 billion, seventh by TVL): a semi-permissioned real-world asset chain without an open DeFi ecosystem where anyone can deploy contracts. Comparing its TVL with Solana's mixes two different things. It is excluded with the reason stated, never in silence.

With this filter, the ten are Ethereum ($40.382 billion), BSC ($4.827), Tron ($4.820), Solana ($4.719), Base ($4.501), Bitcoin ($3.473), Hyperliquid ($1.197), Arbitrum ($1.171), Polygon ($792), and Monad ($752). They total $66.634 billion, of which Ethereum concentrates 60.6% of this top-10 —the 54.3% from our analysis of TVL share by chain measured the total DeFi universe in July, which is larger; hence the difference—. The three-L2 cap did not come into play in this capture: Base, Arbitrum, and Polygon occupy exactly the three spots, and the next relevant L2, OP Mainnet, sits at $299 million, far from the TVL cut-off. The rule is declared in advance regardless, so the next capture doesn't choose it for convenience. Monad enters as tenth with $752 million compared to Plasma's $707 million: a 2026 mainnet with a data history of barely a few months, making its averages much less robust than the rest.

How does the ten-chain scorecard look?

The table crosses the five dimensions. The active wallets column uses the daily average of the last thirty days with the source indicated in the following section; "n/a" cells correspond to chains without a comparable public series and are left empty rather than filled with a figure from a different methodology.

* Tron average is over 15 days, the maximum returned by the Tronscan public API in a single call.

Two cross-readings are immediately apparent. Polygon moves 463,836 active addresses per day with $792 million in deposited capital, while Ethereum moves 487,803 with $40.382 billion: practically the same crowd over fifty times more capital. And Hyperliquid, seventh by TVL on the list, generated $49.4 million in fees in thirty days —more than six times Ethereum's in the same window, $7.7 million— which reorders token return in a way that TVL completely reverses.

Why do two dashboards give different active wallet figures for the same day?

Active wallets is the most fragile of the five metrics, and that fragility is measurable. Three of the ten chains have public series on two independent dashboards simultaneously, allowing for a comparison of the same day, the same network, and the same thirty-day window using two different methodologies.

The Bitcoin gap, nearly 30%, has a mechanical explanation: one dashboard counts unique addresses appearing in any transaction of the day —including change addresses automatically generated by your own wallet when spending— while the other filters out some of that noise. On Ethereum, where accounts are reused, the 16.4% gap mostly stems from whether contract accounts are counted alongside user accounts. Base, with two nearly identical methodologies, sits at 0.5%: the divergence depends on each chain's account model.

Hence the rules for this column. Ethereum, Base, Arbitrum, and Polygon use the daily active addresses series from growthepie, which applies the same criteria to all four. Solana and Bitcoin use The Block. Tron uses Tronscan with a fifteen-day average because its public API does not return more history in a single call. BSC, Hyperliquid, and Monad remain "n/a": BSC and Monad explorers require keys, and Hyperliquid activity is concentrated in an order book outside the EVM account model; for BSC and Monad, there is no verifiable public proxy with a comparable methodology, so the gap is declared. For Hyperliquid, there is a verifiable substitute figure —250,578 addresses with a HYPE balance, according to Hypurrscan on August 3— which measures holding rather than daily activity, so it does not occupy the cell. Given this margin of error, comparing Solana with Tron only makes sense if both figures come from the same dashboard.

What does each chain return to its token: burn, buyback, or nothing?

There are three ways for network activity to return to its token, and they produce results of different orders of magnitude.

Fee-funded burn. Ethereum destroys the base fee of every transaction since EIP-1559 (the 2021 upgrade that introduced it): 4,636,944 ETH accumulated since 2021 (about $13.101 billion at the value of each moment). The current pace, however, has plummeted along with gas prices: 990.6 ETH in the last thirty days, around $1.83 million, which annualizes to roughly $22 million. Polygon applies the same mechanism and offsets it against a 2% annual emission —1% to validators, 1% to the community treasury— so its net balance fluctuates between deflationary and inflationary depending on the month. Monad burns the base component of its fees while emitting 25 MON per block, with the caveat that its historical series barely covers a few months.

Formula-funded burn. The BNB auto-burn removed 1,615,828 BNB on July 15, 2026, about $932 million, in its thirty-sixth quarterly operation, leaving the circulating supply at 133.17 million BNB. The figure is impressive next to Ethereum —about 170 times what the network burns in a quarter— but the comparison breaks down as soon as you look at where the money comes from: the auto-burn calculates the amount based on the BNB price and the number of blocks produced, removing tokens from a reserve until reaching a 100 million supply. The mechanism fueled by actual BNB Chain fees is different, BEP-95 (its automatic per-block burn), which has destroyed about 291,000 BNB since launch. Two numbers on the same chain measuring different things.

Open market buyback. Hyperliquid allocates 97% of protocol fees to its Assistance Fund, which buys HYPE automatically and continuously. The fund's address held 46.19 million HYPE on August 3, 2026, according to Hypurrscan —4.6% of the 1 billion genesis supply— compared to the 45.65 million we recorded in early July. With $49.4 million in fees in thirty days, the annualized buy rate is around $583 million. Details of this model and its protocol-level variants are in our fee switch and buyback comparison.

Nothing. Bitcoin neither burns nor buys back: it emits 3.125 BTC per block and fees go entirely to the miner. Arbitrum had no mechanism until the July 9, 2026 announcement, which directs 100% of Arbitrum One fees to the treasury controlled by ARB holders —a treasury, not a distribution. Base has no token to remunerate. And Solana reduced its own return in July 2026: after the approval of SIMD-0096 (the Solana improvement proposal approved in 2024), 100% of priority fees go to the validator that includes the transaction, and only half of the base fee continues to be burned—the small part of a $14.8 million bill over thirty days. The trajectory of what happens to a token when that backing arrives late is documented in our tracking of HYPE, UNI, and ARB airdrops one year later.

What is the native token for if you stop speculating with it?

The honest question for any base-layer token is what forces you to hold it if the secondary market disappeared tomorrow. Five functions exhaust the catalog and serve as a reusable framework for any chain.

Scoring each chain in these five boxes reorders the list again. HYPE, POL, and MON check four of the five. ETH also checks four, with the caveat that its fee capture box is currently worth about $22 million annually. BTC checks two, gas and collateral, and its lack of staking and capture is by design, not a sudden deficiency. And ARB checks only one: Arbitrum gas is paid in ETH, the token backs no consensus, and until July 2026, it received no flow. A token that only votes is exactly what it claims to be, and the market takes years to discount the difference.

Base deserves its own blank row: the fifth chain by TVL operates with $4.501 billion deposited and 276,352 daily active addresses without a native token. Its sequencer fees —$59.7 million in the last twelve months— go to the operator and the Optimism Collective. Anyone wanting exposure to its growth must buy something else, and that gap is the recurring argument for a token that, as of August 2026, remains unconfirmed. If you need to review what TVL, fees, or staking mean, our DeFi metrics guide explains them one by one.

Who really decides in each of these chains?

The fifth dimension is measured by the Nakamoto Coefficient: the minimum number of entities that, by agreeing, would control the network. The lower the number, the more concentrated the power. Nakaflow publishes updated figures —the capture on August 3, 2026, at 06:04 GMT gives Monad 21, Solana 10, BSC 7, Polygon 4, Hyperliquid 3, and Base 1— and the four missing chains can be calculated manually with public data.

For Bitcoin, the equivalent is mining concentration. In the 4,369 blocks of the last month, Foundry USA signed 26.5%, AntPool 18.5%, and F2Pool 16.1%: three pools account for 61.0% of the effective hashrate, according to mempool.space on August 3. Bitcoin appears as "No DAO" in the table because it lacks on-chain governance —rules are changed by social consensus among developers, miners, and nodes— which doesn't prevent measuring where the block production capacity accumulates.

Tron is the opposite case: it has formal and highly concentrated on-chain governance. Its Super Representatives are elected by freezing TRX to vote, and the live count on August 3 distributes 44.173 billion votes among 439 candidates. The top nine —led by Poloniex with 11.3%— accumulate 53.5%, so nine entities are enough to decide. The complementary data is participation: those 44.173 billion votes equal 46.6% of the 94.891 billion TRX in circulation, a very high involvement compared to almost any DAO.

Ethereum also has no DAO: changes go through the EIP process and voluntary client adoption. Its real concentration lives in staking, where the verifiable reference is that Lido holds $17.3 billion of the $27.3 billion in Ethereum liquid staking, 63.4% of that segment according to DefiLlama on August 3. Hyperliquid combines the lowest Nakamoto Coefficient on the list after Base —three validators— with a protocol fund holding 4.6% of the supply. And Arbitrum, the only classic DAO in the group, delegates voting to a handful of addresses; the exact breakdown of the distribution requires paid access to the Tally dashboard, so it is the only row in the group without a quantified coefficient.

How to apply this scorecard to a chain not on the list?

The value of the framework is that it can be replicated with free data in less than an hour. The workflow:

• TVL and fees, from the DefiLlama /v2/chains and /summary/fees/<chain> endpoints. Always note the capture date alongside the figure.
• Active wallets, from a single dashboard for all chains you compare. If you must mix sources, write which one you used in each row and expect a gap of up to 30% between methodologies.
• Token return: first identify if the burn or buyback is funded by fees or a formula based on reserves. A reserve-fueled mechanism eventually runs out; a fee-fueled one scales with usage.
• Marginal utility: run the token through the five boxes —gas, security, fee capture, voting, collateral— and count how many it actually checks, not how many the whitepaper promises.
• Concentration: Nakamoto Coefficient if published; if not, the share of the top block producers or top voters, calculated from the chain explorer.

The five columns purposely disagree with each other. Solana leads in activity and its token return just decreased; Hyperliquid buys back at the highest rate on the list with three validators deciding; Base attracts $4.501 billion without a token and with a single sequencer; Bitcoin returns nothing to its token. No one wins in all five, and that is the reason for scoring them separately instead of looking for a summary number. The next capture will show which cells moved and which have been still for two years, which is the information that no monthly ranking preserves.

Sources and links: DefiLlama — TVL and fees by chain · growthepie — daily active addresses · The Block — active addresses on Solana · Blockchain.com — Bitcoin unique addresses · ultrasound.money — ETH burn · Crypto Briefing — 36th BNB quarterly burn · Hypurrscan — HYPE holders and Assistance Fund · Tronscan — Super Representative votes · mempool.space — mining pool distribution · Nakaflow — Nakamoto coefficients · Solana Docs — fee structure · Monad — MON tokenomics