Ethereum in 2026 is displaying an extreme structural divergence: network utility keeps climbing, yet the token's value is facing a severe test.
The ETH/BTC ratio has slid to 0.029, only about 4% above its late-October low, while leading holders such as BitMine (BMNR.US) are replacing direct coin purchases with share buybacks, marking a fundamental shift in traditional capital logic.
This rupture between "useful" and "valuable" forms the core narrative conflict in the current market.
Traditional Capital Retreat and Corporate Strategy Shift: From Buying Coins to Buybacks
Signs of capital outflow on the macro level became especially pronounced in October. U.S. spot Ethereum ETFs saw net outflows exceeding $1.6 billion that month, with the ETHA fund under BlackRock (BLK.US) becoming the main channel of redemption.
Looking back since January 2026, funds showed net inflows in only two months, while the remaining months were marked by continuous outflows. The deeper reason lies in regulatory constraints: ETF products cannot participate in staking, so holders cannot earn yield.
In a dollar interest-rate environment, the risk-free yield offered by U.S. Treasuries is more attractive to institutions seeking absolute returns, leaving investors unwilling to hold a fund that only loses and never earns.
At the same time, corporate treasury accumulation strategies have also changed markedly. BitMine, the world's largest corporate Ethereum holder, currently holds about 6.06 million ETH, nearly 5% of the circulating supply, approaching its self-imposed holding cap.
Most of these assets are locked in staking to earn interest, but its average cost is around $3,500, while the current market price hovers near $2,300, resulting in unrealized losses of about $7.5 billion, or roughly 35%.
Second-ranked SharpLink (SBET.US) holds about 889,000 ETH and is likewise mired in unrealized losses. Together, the two companies hold close to 6% of the circulating supply.
Management has shown extreme rationality in financial decisions. Because BitMine's share price is only 0.88 times net asset value, a significant discount, using staking income to repurchase discounted shares is more economically efficient than continuing to add ETH at high prices in the market.
Year to date, BitMine has spent close to $400 million on share buybacks, far more than it spent buying ETH over the same period. Although the company is still buying ETH, the pace has clearly slowed.
While financial optimization at the micro level is reasonable, the spot market's largest potential buyer is pulling back, which materially weakens price support.
On-Chain Capital Holds Firm: Staking Queues and a Security Stress Test
In sharp contrast to the coldness of traditional markets, on-chain data shows strong lock-up demand. In late October, about 1.37 million ETH were queued for staking activation, with an expected wait of more than 24 days; by comparison, only about 13,000 ETH were queued for exit, a process that takes just over a day.
This "hard to enter, easy to leave" situation shows that entry demand far exceeds the network's current processing capacity. Even though the annualized yield has been diluted to 2.61%, capital still chooses to give up liquidity in exchange for on-chain interest, demonstrating the firm conviction of long-term holders.
The Ethereum protocol processes at most about 57,600 ETH of entries and exits per day, and a queue of 1.37 million ETH highlights the supply-demand imbalance.
In addition, the network underwent a stress test in early October: MetaMask voluntarily exited about 17,000 validators due to a security incident, briefly pushing the exit queue to about 550,000.
Yet the network absorbed this shock smoothly within three weeks, with total staked ETH holding steady above 43.7 million. This shows it was not a panicked flight but a temporary fluctuation caused by a security event.
The network successfully swallowed the impact without systemic risk, further verifying the resilience of its underlying architecture.
The Double-Edged Sword of Collapsing Fees: ENS Returns to Mainnet and Deflation Stalls
Data compiled by Woofun AI shows that Ethereum mainnet fees fell by more than 60% year over year in the third quarter. In October, the average fee for a single mainnet transaction was only $0.25.
Back in the network congestion period of 2021, the same operation often cost more than $53. The fee for one ETH transfer three years ago is now enough to cover more than two hundred transfers.
This ultra-low-cost environment has directly changed application-layer deployment strategies. The Ethereum Name Service, ENS, had originally planned to build a dedicated layer-2 network, NameChain, to migrate domain registration and resolution to L2.
However, in February 2026, the team announced it was abandoning that route and decided to deploy ENS v2 entirely on mainnet. With mainnet interaction costs falling to a few cents, the operating costs of building one's own L2, the security risks of cross-chain bridges, and the problem of liquidity fragmentation no longer seemed worth it.
Staying on mainnet is not only cheaper but also lets the domain system compose directly with lending, trading, and identity protocols without waiting for cross-chain confirmations. ENS became the first major protocol to abandon L2 because mainnet became cheap, a trend that could bring more applications back.
Yet low fees dealt a fatal blow to the token model. Ethereum's automatic burn mechanism depends on the transaction base fee; the higher the fee, the greater the burn. The base fee is currently only 0.06 Gwei, while achieving burn-issuance balance requires 13.63 Gwei, a gap of two orders of magnitude.
As a result, the network is issuing a net increase of about 2,954 ETH per day, creating nearly $7 million of sell pressure out of thin air each day at current prices.
The "ultrasound money" deflation narrative that was so popular in 2021 has completely stalled: the busier the network, the more inflationary the token, and the value transmission chain is broken.
The L2 Narrative Recedes and Leading Protocols Contract
As mainnet costs fall, the necessity of L2 is being questioned. In February 2026, Vitalik Buterin wrote that branded sharding "no longer makes sense."
He criticized many L2s for claiming high TPS while relying on multisig wallets managed by a few people for asset transfers, essentially centralized external databases, with slow progress on decentralization and permissionless fraud proofs still not delivered.
When mainnet itself becomes cheap and roomy, the rationale for L2s lacking independent security guarantees is greatly weakened. A market shakeout followed.
Kroma stopped service, Kinto shut down entirely, and Blast's TVL plunged 97% after the airdrop wave receded. Leading protocols are also pulling back: besides ENS returning to mainnet, Synthetix and prediction market Trueo scaled back deployments on long-tail L2s, refocusing on mainnet.
Although Aave has not fully migrated back, it is gradually shutting down small deployments lacking activity and concentrating core liquidity on mainnet and a few leading layer-2s. The direction is clear: those that can return to mainnet do so, and those that cannot contract.
The dividend of low fees and the cost of stalled deflation ultimately point to the same contradiction.
Upgrade Stalled and the State of the Ecosystem Moat
Ethereum's next major upgrade, Glamsterdam, aims to raise the block gas limit from 60 million to 200 million, tripling mainnet throughput or more.
The upgrade activated smoothly on the Sepolia testnet on October 6, but the plan to launch on the Hoodi testnet on October 26 was urgently halted, with no new timetable announced.
The obstacle came from a security audit report released by ChainSecurity on October 13, which pointed out that EIP-8037 in the upgrade changed how the virtual machine handles failed transactions, potentially exposing 13 contracts already running on mainnet to vulnerabilities.
These contracts were supposed to automatically roll back when transactions fail to protect user funds, but the new rules changed the resource refund mechanism and could render that protection ineffective. Because blockchain code is immutable, the development team cannot patch third-party contracts and can only modify the proposal itself.
The Hoodi delay means the December mainnet upgrade window is closed, and the 200 million gas target has lost its timetable. Ethereum chose safety over speed.
Despite the stalled upgrade, Ethereum still has a deep moat: mainnet carries more than half of the network's stablecoin supply and over $20 billion in tokenized real-world assets, with Treasury products from BlackRock and Franklin Templeton (BEN.US) both running on it.
However, stablecoin market value also bled in the third quarter, with erosion by low-cost networks such as Tron clearly visible. Vitalik said the foundation will become "a smaller ship," focusing on security and core research and development while reducing direct ETH sales.
The network is getting better, but the tension between the token model and security constraints continues to intensify.
Comments