48 posts tagged with "Staking"

A single public company now controls roughly 3.5% of every ETH ever issued, and 82.59% of that hoard is actively earning validator yield. On May 2, 2026, wallets tied to BitMine Immersion Technologies (NYSE: BMNR) deposited another 162,088 ETH — about $366 million at spot — into Coinbase Prime staking contracts, lifting the company's total staked position to 4,194,029 ETH worth $9.48 billion. The number that matters is not the dollar figure. It is the ratio.

Most ETH treasury vehicles run a staking ratio of zero. ETF wrappers are barred from staking under current SEC structure, MicroStrategy-clone copycats default to passive cold storage, and even Coinbase Custody clients spread their ETH across many third-party operators. BitMine's 82.59% staked ratio is the most aggressive validator-yield treasury strategy in public markets, and it forces a reset on what an "ETH treasury company" actually is. This is no longer a passive accumulation play. It is a publicly traded validator company.

The May 2 Deposit and the Math Behind 82.59%​

The transaction itself was almost routine: a Coinbase Prime staking deposit eight hours after BitMine's prior buys settled, routed through MAVAN — the company's proprietary validator network launched March 25, 2026. What was not routine was the cumulative effect. With 4,194,029 ETH now staked, BitMine alone is responsible for roughly 11% of all staked Ethereum supply, a tier previously reserved for protocols like Lido (which still controls 23-28.5% of staked ETH across thousands of node operators) and Coinbase Custody (which intermediates for many institutional clients).

At today's blended 3.3% network APY — and closer to 5.69% for validators that fully participate in MEV-Boost — BitMine's annualized staking revenue lands somewhere between $260 million and $360 million. That is more than the entire net income of many mid-cap fintech listings. It is also a recurring, on-chain, ETH-denominated cash flow that compounds back into the position itself.

The 82.59% number deserves scrutiny because it implies an operational discipline most ETH treasuries lack:

• The remaining 17.41% sits unstaked as a liquidity buffer, presumably reserved for working capital, Treasury management, and the next round of buys before they are routed into validators.
• Onboarding 162,088 ETH in a single deposit means BitMine is comfortable absorbing the activation queue delay (which spiked to 45 days at peaks earlier in 2026) rather than waiting for spot purchases to clear before staking.
• The company is effectively saying: every dollar of marginal ETH should produce yield, and unstaked balances are a drag, not a feature.

Compare that to Strategy (formerly MicroStrategy), which holds roughly $71 billion in Bitcoin but earns zero yield on the position. Strategy's playbook depends entirely on price appreciation. BitMine's playbook layers a 3-5% native yield on top of price appreciation — a structurally different return profile that turns ETH into something closer to a tokenized perpetual bond than a digital commodity.

The ETH Treasury Race Has a New Top Tier​

Before BitMine's pivot from Bitcoin mining to an Ethereum-treasury strategy, the ETH treasury company category was a curiosity. SharpLink Gaming (SBET) — once on the brink of delisting — reinvented itself as "the Ethereum MicroStrategy" and built a roughly 868,699 ETH position by early 2026. The Ether Machine (ETHM) sits at around 496,712 ETH. Bit Digital (BTBT) holds about 155,444 ETH. Coinbase carries ETH on its corporate balance sheet as part of operational reserves.

BitMine eclipses all of them combined.

Company	ETH Holdings (approx.)	Staking Posture
BitMine Immersion (BMNR)	~4.97M ETH	82.59% staked via MAVAN
SharpLink Gaming (SBET)	~869K ETH	Partial staking, third-party operators
The Ether Machine (ETHM)	~497K ETH	Mixed
Bit Digital (BTBT)	~155K ETH	Limited

The gap is not just about scale. BitMine's stated target is 5% of all ETH issuance. At current pace, the company is roughly 81% of the way to that goal. If it gets there — and the May 2 deposit suggests management considers it a question of when, not if — a single Nasdaq-listed entity would hold a sovereign-tier ETH position.

That changes the negotiation. ETH treasury companies of this scale do not buy spot from open-market exchanges; they call the Ethereum Foundation, OTC desks, and large stakers directly. Recent reporting confirms BitMine has acquired ETH directly from the Ethereum Foundation in tranches totaling tens of millions of dollars — the Foundation is, in effect, recycling treasury sales into the largest single-company validator on its own network.

MAVAN: From Treasury Tool to Infrastructure Business​

The Made in America Validator Network was originally built for one customer: BitMine itself. Its purpose was to give the company sovereign control over its validators rather than relying on Figment, Kiln, Anchorage, or Coinbase Cloud. By March 25, 2026, MAVAN was running roughly $6.8 billion in ETH on US-based infrastructure with a globally distributed architecture for institutional clients who want non-US validation.

Two strategic moves separate MAVAN from the dozens of other staking-as-a-service products:

1. It plans to externalize. BitMine has signaled MAVAN will sell staking services to institutional investors, custodians, and ecosystem partners — turning the validator stack from a cost center into a revenue line. This is the same playbook AWS ran when it externalized Amazon's internal infrastructure in 2006: build something you need anyway, then sell the surplus.

2. It is multi-chain. BitMine projects MAVAN expanding beyond Ethereum to additional proof-of-stake networks during 2026. The economics suggest validator infrastructure for chains like Solana, Sui, Aptos, and Cosmos-aligned networks could rival or exceed Ethereum staking margins, especially as those chains attract institutional capital.

The financial implication is that BMNR is no longer just a leveraged ETH play. It is a leveraged ETH play plus a staking infrastructure business with margin compounding across multiple PoS networks. Investors trying to value the stock as "ETH ÷ shares outstanding" are missing the second leg.

The Centralization Question Nobody Wants to Ask​

Concentrating 11% of staked ETH in a single corporate entity raises a question Ethereum's social layer has historically tried to avoid: what does decentralization mean when the largest validator operator is a US-listed public company subject to OFAC, FinCEN, and SEC oversight?

The technical risks are well-rehearsed:

• A single entity controlling >33% of staked ETH could theoretically delay finality. BitMine alone is far below this, but combined with other US-regulated stakers (Coinbase, Kraken, Figment, Anchorage), the addressable concentration risk grows.
• Compliance pressure could force MAVAN validators to censor transactions matching OFAC lists, replaying the 2022-2023 MEV-Boost relay debate at a much larger scale.
• Slashing events, infrastructure outages, or regulatory action against BitMine could remove validators with material network impact.

Ethereum's response options are limited. EIP-7251 (max effective balance increase to 2,048 ETH) reduces the number of validators a large staker needs to run, which arguably concentrates control further by making consolidation cheaper. Distributed validator technology (DVT) promises to spread key control across multiple node operators without changing economic ownership, but adoption remains nascent. Liquid staking protocols like Lido have introduced Community Staking Modules to broaden their operator base — but Lido's roughly 23-28.5% share is itself the second-order centralization concern.

The honest framing: Ethereum's economic decentralization is migrating from a long tail of solo stakers to a handful of institutional operators with very different incentive structures. BitMine's MAVAN, Lido's CSM, BlackRock's staking-enabled ETF posture, and Grayscale's 1.16M ETH January staking deposit all push in the same direction — institutional dominance of the validator set.

That migration may be inevitable. It is not necessarily catastrophic. But pretending it is not happening because BitMine "only" runs 11% of staked supply ignores how the numbers compound.

Supply Compression Meets Staking Demand​

The May 2 deposit also matters because of where Ethereum's supply curve sits in mid-2026. With BitMine staking 4.19M ETH and the broader ecosystem locking up roughly 35.86M ETH (28.91% of total supply), circulating float is materially tighter than the headline market cap suggests.

Layer in three forces actively compressing supply through 2026:

• Ethereum Foundation's Treasury Staking Initiative committed 70,000 ETH to direct staking starting February 2026, with rewards looped back into the EF treasury.
• Staking-enabled ETFs now represent over 40% of institutional Ethereum investments, pulling float out of exchanges and into long-duration custody.
• Validator entry queues hit 2.6 million ETH at peaks earlier in 2026, with 45-day activation waits that incentivize early deposits.

When 82% of a $11.5 billion treasury chooses to disappear into 32-ETH validator commitments, that is structural sell-side absorption. Anyone modeling ETH's 2026 supply-demand needs to treat BitMine's behavior as a price-insensitive bid until management says otherwise.

What Comes Next​

The interesting question is whether the BitMine model triggers imitation. Three scenarios are plausible by year-end 2026:

1. Imitation accelerates. SharpLink, The Ether Machine, and a wave of new SPAC-listed ETH treasury vehicles raise capital specifically to run their own validator networks. Multi-chain staking infrastructure becomes the default treasury structure, and "ETH treasury company without proprietary validators" becomes the underperforming category.

2. Regulatory friction caps it. SEC, FASB, or OFAC guidance treats staking revenue as activity income subject to additional disclosure, audit, or capital requirements. Public-company economics deteriorate enough that managers default back to passive holding, ceding the validator economy to private operators and protocols.

3. Decentralization pressure forces fragmentation. Ethereum's social layer (or a coordinated set of solo stakers and DVT advocates) successfully pushes BitMine and peers to distribute key control across multiple operators rather than running unified internal infrastructure. The economics survive but the validator topology flattens.

The May 2 transaction does not resolve any of those scenarios. It does ratify one fact: validator yield is no longer optional for a competitive ETH treasury, and the largest player just lapped the rest of the field.

BlockEden.xyz provides enterprise-grade Ethereum RPC and staking infrastructure for builders running across 30+ chains. Explore our API marketplace to plug your validator dashboards, treasury tooling, and on-chain analytics into infrastructure designed for institutional load.

Sources​

• Bitmine Doubles Down on Ethereum with Massive $366M Staking Deposit — CryptoTimes
• BitMine's 4.5M ETH stake strategy boosts confidence in $5,000 May target — Crypto Briefing
• Bitmine Crosses 10% of ETH Staking Supply — Coinpedia
• BitMine's ether treasury hits 4.8 million ETH as stock listing moves to NYSE — CoinDesk
• Bitmine Immersion Technologies (BMNR) Announces Launch of MAVAN — PR Newswire
• BitMine Launches Proprietary Ethereum Validator Network MAVAN — The Defiant
• Ethereum Treasuries: ETH Holdings of Public Companies & Governments — CoinGecko
• Ethereum Staking Statistics & Trends in 2026 — Datawallet
• Staking for Treasuries: How Institutions Allocate to PoS Networks — Everstake
• Ethereum Foundation starts 70,000 ETH staking process — CoinDesk
• Bitmine Amasses 4.1M ETH: Staking Giant or Decentralization Risk? — Let's Talk Bitcoin
• Ethereum Validator Queue — validatorqueue.com

On April 30, 2026, Telegram staked 2.2 million TON — roughly $2.88 million at the time — and switched on as a primary validator on The Open Network. The headline number is almost a rounding error in crypto. The signal underneath it is not.

For the first time, a consumer platform with 950 million monthly active users is not just partnered with a Layer 1 — it is helping secure it, propose blocks on it, and finalize transactions on it. Pair that with the Catchain 2.0 mainnet upgrade that just collapsed TON's block time from 2.5 seconds to 400 milliseconds, and a 6x fee cut to a flat $0.0005 per transaction, and a different kind of question starts to come into focus. TON is no longer trying to beat Solana on TPS or Ethereum on TVL. It is starting to look like an attempt to compete with WeChat Pay, Apple Pay, and Stripe — using a blockchain as the rail.

A market maker became an asset manager last week, and almost nobody noticed.

On April 22, 2026, GSR — the 13-year-old institutional liquidity firm best known for OTC desks and a landmark confidential trade on encrypted Ethereum — listed the GSR Crypto Core3 ETF on Nasdaq under the ticker BESO. The fund holds Bitcoin, Ether, and Solana in actively-managed proportions, rebalances weekly off proprietary research signals, and — critically — pockets staking yield on the ETH and SOL sleeves. It is the first U.S.-listed multi-asset crypto ETF authorized to stake.

That last sentence is doing a lot of work. For two years, the question hanging over every spot-ETF approval was whether the SEC would ever let issuers earn the on-chain yield that distinguishes a productive asset from inert digital gold. The answer, finally, is yes. And the firm cashing the first check is not BlackRock, not Fidelity, not Bitwise. It's a market maker that, until last week, didn't run a single dollar of public fund AUM.

In a single twelve-day window, two general-purpose Layer 1s reached for the same number — 2.1 billion. On March 12, 2026, Polkadot activated a hard cap of 2.1B DOT through Referenda #1710 and #1828. On April 14, Aptos governance passed Proposal #183 with 335.2 million APT in favor and just 1,500 opposed, locking the same 2.1B ceiling on APT supply alongside a 50% staking-yield cut and 100% gas-fee burn. The numerical coincidence is not what matters. The signal is.

For three years, the prevailing alt-L1 playbook treated supply expansion as a feature: emissions funded validator security, ecosystem grants subsidized developer adoption, and the assumption was that demand would eventually outrun dilution. In 2026, that assumption is being abandoned in real time. Aptos, Polkadot, and a growing list of competitors are converging on a Bitcoin-shaped narrative — capped float, fee burns, foundation-locked tokens — at exactly the moment Solana's uncapped model becomes the loudest outlier in the room.

For seventeen years, Bitcoin's defining feature was that it did nothing. You bought it, you held it, you waited. The asset that birthed an entire industry was, paradoxically, the only major one that couldn't participate in it. As of April 2026, less than 1% of Bitcoin's circulating supply is locked in any form of DeFi — a stunning statistic when you consider that BTC alone represents roughly $1.9 trillion of capital sitting still while $7 billion of "Bitcoin DeFi" tries to wake it up.

That gap is the largest unallocated yield opportunity in crypto. And four very different protocols — Babylon, Stacks' sBTC, Threshold's tBTC, and exSat — are racing to define how Bitcoin becomes programmable collateral without forcing holders to trust a custodian, abandon the base chain, or lose the property that made them buy BTC in the first place: that nobody can take it away.

This is the Bitcoin-backed stablecoin economy of 2026. It is messier, more contested, and far more strategically important than the wrapped-BTC story Wall Street tells.

When a company buys $233 million worth of ether in seven days and barely makes a headline, you know the corporate crypto treasury arms race has officially crossed into a new phase. That is exactly what happened the week ending April 22, 2026, when Bitmine Immersion Technologies (BMNR) added 101,627 ETH — its largest single-week accumulation of the year — to push total holdings past 4.98 million tokens. By the company's April 27 update, that figure had climbed again to 5.078 million ETH and roughly $13.3 billion in total crypto and cash on the balance sheet.

Tom Lee's bet is no longer a curiosity. It is the most aggressive corporate treasury experiment in Ethereum's history, and it is starting to look like a structural mirror of Michael Saylor's Bitcoin playbook — only with a yield engine bolted on. The question for the rest of 2026 is whether the Bitmine model creates a stable new class of public-market ETH proxy, or whether the same reflexive dynamics that made Strategy a $63 billion juggernaut also seed the next forced-seller cascade.

In March 2026, a few dozen anonymous miners on home internet connections trained a 72-billion-parameter language model that scored within striking distance of Meta's Llama 2 70B. Six weeks later, the team that led that effort walked out, dumped $10 million worth of TAO, and called Bittensor's decentralization "theatre." Now the surviving community wants to do it again — at fourteen times the scale, in roughly four weeks, with the entire decentralized AI thesis riding on the result.

This is the story of how Bittensor's Subnet 3 — recently rebranded Teutonic after the Covenant AI exit — talked itself into a 1-trillion-parameter training run timed to land squarely in Grayscale's TAO ETF SEC review window. It's a wager that the protocol's incentive layer is more important than the people who built it, and that the same network that survived a governance crisis can ship the "DeepSeek moment" for decentralized AI before regulators decide whether to let Wall Street buy in.

How a 72B model became the high-water mark for permissionless AI​

The story starts on March 10, 2026, when Subnet 3 — then operating under the name Templar — announced Covenant-72B, a 72-billion-parameter model trained on roughly 1.1 trillion tokens by more than 70 independent miners coordinating across the public internet. It was, by a wide margin, the largest decentralized LLM pre-training run ever completed.

The benchmark that mattered: an MMLU score of 67.1, putting Covenant-72B in the same neighborhood as Meta's Llama 2 70B — a model produced by one of the best-funded AI labs on the planet. NVIDIA CEO Jensen Huang publicly compared the effort to a "modern folding@home for AI." Templar's subnet token surged, and at peak its market valuation crossed $1.5 billion.

The technical breakthrough wasn't the model architecture. It was the coordination layer. Two pieces did the heavy lifting:

• SparseLoCo, a communication-efficient training algorithm that reduced inter-node bandwidth requirements by 146x through sparsification, 2-bit quantization, and error feedback. Without it, a frontier-scale training run on residential internet would be physically impossible — gradient sync alone would saturate every miner's connection.
• Gauntlet, Bittensor's blockchain-validated incentive system that scored each miner's contribution via loss evaluation and OpenSkill rankings, paying TAO to the high-quality nodes and slashing the rest.

Together they produced something genuinely new: a permissionless network of anonymous contributors, coordinating only through cryptographic incentives, training a model competitive with billion-dollar lab outputs.

Then it broke.

The Covenant exit: $900 million erased in twelve hours​

On April 10, 2026, Sam Dare — founder of Covenant AI, the team behind three of Bittensor's most valuable subnets (SN3 Templar, SN39 Basilica, and SN81 Grail) — announced he was leaving. Within hours he liquidated approximately 37,000 TAO, roughly $10.2 million, and published a parting accusation: that co-founder Jacob Steeves ("Const") wielded centralized control over the protocol, and that Bittensor's decentralization was performance, not architecture.

The market reaction was immediate. TAO crashed 20–28% depending on the measurement window, erasing roughly $650–900 million in market cap inside a 12-hour span. Subnet alpha tokens fared worse — Grail (SN81) was down 67% at the bottom. Around $10 million in long positions liquidated.

Two facts blunted the panic:

1. The subnets didn't die. Community miners restarted SN3, SN39, and SN81 from open-source code without a central operator. The infrastructure Covenant built was, in fact, recoverable from the public artifacts — which arguably proves the decentralization thesis Dare disputed.
2. 70% of TAO supply remained staked through the disruption. Long-term holders didn't follow Dare to the exit.

But the network had a credibility problem. If Covenant — the team that delivered Bittensor's marquee technical achievement — could leave at the top and crater the token, what stops the next subnet operator from doing the same?

The Conviction Mechanism: locking in the people who can leave​

Const's response landed on April 20, 2026, ten days after Dare walked. BIT-0011, branded the Conviction Mechanism, proposes a Locked Stake regime that forces subnet owners to time-lock TAO for months or years in exchange for a "conviction score" that maps to voting rights and subnet ownership.

The mechanics:

• The conviction score starts at 100% and decays over 30-day intervals if tokens aren't replenished into the lock-up.
• Voting power and ownership rights diminish in lockstep with the decay, making sudden capital flight economically expensive rather than just embarrassing.
• The system targets the mature subnets first — SN3, SN39, and SN81 — exactly the three that Covenant ran.

The dark joke: BIT-0011 was reportedly drafted by Sam Dare himself before his exit. The departing founder wrote the rules designed to prevent founders from departing.

The proposal addresses a real structural weakness — subnet operators could previously dump positions with no governance penalty — but it also concentrates power in the hands of long-term lockers, which is its own form of centralization. Whether that's the right trade depends on what you think Bittensor's main risk is: founder defection or oligarchic capture.

Teutonic and the trillion-parameter moonshot​

Against that backdrop, the rebranded Teutonic subnet (SN3, formerly Templar) has committed publicly to a 1-trillion-parameter decentralized training run for mid-to-late May 2026. That's roughly 14x the scale of Covenant-72B, on the same fundamental architecture, with a community-restored team rather than the original Covenant engineers.

The strategic timing is impossible to miss. Grayscale filed its S-1 amendment for the spot Bittensor Trust ETF (proposed ticker GTAO) on NYSE Arca on April 2, 2026. The SEC's decision window is currently tracked for August 2026. A successful 1T-parameter training run in May would land at the peak of regulator deliberation — exactly when "is this a real technology or a meme?" becomes the load-bearing question. Grayscale already raised TAO's weighting inside its broader AI fund to 43.06% on April 7, the largest single-asset reallocation that fund has ever made.

The bull case writes itself: ship a credible 1T-parameter decentralized model, become the "DeepSeek moment" the ETF approval needs to justify institutional inflow, and reprice the entire decentralized AI category in one quarter.

The bear case is engineering, not marketing.

Why scaling decentralized training is hard in ways frontier labs don't face​

Centralized 1T+ models — GPT-5, Claude 4.7 Opus, Gemini 2.5 Ultra — are trained inside facilities where every GPU is wired to every other GPU through purpose-built fabrics like NVLink and InfiniBand, with sub-microsecond latencies and terabit-per-second bandwidth. Even in those conditions, gradient synchronization is the bottleneck. Published research consistently finds that over 90% of LLM training time can be spent on communication rather than compute when scaling is naive.

Teutonic's miners are coordinating across ~100ms WAN latencies on residential internet. The only reason Covenant-72B was possible at all is SparseLoCo's 146x compression of communication volume. Pushing to 1T parameters changes the math in three uncomfortable ways:

1. Gradient size scales roughly linearly with parameter count. A 14x model means 14x as much data to synchronize per step, even before considering optimizer state.
2. Cross-node coordination overhead historically scales super-linearly with worker count. If Teutonic doubles its node pool from ~70 to ~256, the all-reduce communication cost doesn't just double — it can grow by 4–10x depending on topology.
3. Failure modes compound. A node dropping out mid-step in a 70-node network is a small slashing event. In a 256-node network running 14x larger gradients, the same drop can stall the entire training round.

None of this is unsolvable. There's a body of decentralized training research — heterogeneous low-bandwidth pre-training, FusionLLM, communication-computation overlap, delayed gradient compensation — that targets exactly this regime. But almost all of it has been validated at the 7B–70B scale. A 1T-parameter run on geographically distributed commodity hardware would be a research contribution in its own right, not just a product launch.

The honest read: Teutonic is taking on a research-grade engineering challenge with a marketing-grade deadline. Either it works and becomes the credibility event the entire dTAO ecosystem needs, or it stalls publicly during the SEC's most attentive review window.

The decentralized AI training landscape Teutonic must survive​

Teutonic isn't the only project trying to claim the "credible decentralized 1T-param" milestone in 2026. The competitive map is filling out fast:

• Gensyn launched its mainnet on April 22, 2026 — the same day this article goes out — pairing the launch with Delphi Markets, an AI-driven matching layer for compute jobs. By close of day Gensyn was reporting hashrate equivalent to 5,000+ NVIDIA H100s. Where Bittensor sells permissionless coordination plus a token-incentive flywheel, Gensyn is positioning as a verifiable AI compute marketplace with cryptographic proofs of correct execution.
• Ritual has gone in the opposite direction, leaning into inference rather than training. Its Infernet technology lets any smart contract request an AI output and receive cryptographic proof that the specified model was used unmodified. That's the "verifiable AI in DeFi" thesis, not the "train frontier models from scratch" thesis.
• Ambient and Origins Network are making adjacent bets — different incentive designs, different verification strategies, similar long-term goal of breaking centralized labs' monopoly on frontier training.

These projects don't directly compete on the same milestone, but they all compete for the same finite pool of attention and capital. If Gensyn's mainnet captures the "decentralized AI is here" narrative through commercial workloads, Teutonic's May training run becomes a referendum on whether Bittensor's specific approach — subnet competition plus token-weighted incentives — is the right architecture or the first iteration that gets surpassed.

Why this matters beyond TAO​

Three things are getting tested simultaneously over the next four to six weeks:

Whether decentralized training scales. If Teutonic succeeds, the "Bitcoin of decentralized AI compute" thesis survives. If it fails, the Covenant exit reads as the moment subnet-based training peaked — a 72B ceiling rather than a 72B foundation.

Whether the Conviction Mechanism is the right governance fix. Locking in subnet operators prevents another Covenant-style dump but creates a new failure mode where long-term lockers can entrench. Bitcoin Core's distributed maintainer model, Solana Labs' continued centralized core development, and Sui's Mysten Labs concentration are three different answers to the same question — whether protocol complexity demands a strong central maintainer the community must trust. Bittensor is now running its own version of that experiment in real time.

Whether the ETF window forces decentralized AI to ship on TradFi's calendar. The SEC's August decision window is a hard deadline for a narrative that wants to be "DeepSeek moment" rather than "interesting research project." That's a healthy forcing function or a recipe for over-promising — depending on what gets shipped.

For builders watching from the infrastructure side, the underlying signal is simpler: AI agents and decentralized training networks are about to generate a new tier of on-chain query load — model registry lookups, attestation proofs, gradient checkpoint hashes, subnet performance data — that doesn't fit neatly into the human-facing dApp pattern existing RPC infrastructure was built for.

BlockEden.xyz provides enterprise-grade RPC and indexing infrastructure across 27+ chains for teams building the AI-meets-crypto stack. Explore our API marketplace to build on rails designed for both human and machine traffic.

Sources​

• Bittensor's SN3 Completes Largest-Decentralized LLM Pretraining Run in History — AInvest
• Bittensor's Templar Subnet Completes First Frontier-Scale Decentralized LLM Training — 800.works
• Bittensor Covenant-72B Explained — Phemex
• TAO's DeepSeek Moment: The Rise of Templar (SN3) — PANews
• Covenant AI's Bittensor Exit Triggers 23% TAO Price Crash — Crypto Times
• Grayscale Files Spot TAO ETF as Bittensor Network Rebounds from Covenant AI Exit and 38% Drawdown — Blockonomi
• Bittensor TAO Proposes Locked Stake Governance Fix Following Covenant AI Exit — AInvest
• Bittensor TAO Governance Crisis Explained: Covenant AI Exit and BIT-0011 Proposal — AInvest
• Gensyn launches its mainnet — Startup Fortune
• Beyond A Single AI Cluster: A Survey of Decentralized LLM Training

For every dollar stolen from KelpDAO on April 18, 2026, another $45 walked out of DeFi. That is the ratio the post-mortems keep returning to — a $292 million exploit that detonated into a $13-14 billion TVL exodus in two days, dragged the entire DeFi sector to its lowest total value locked in a year, and convinced a growing share of the institutional buyside that "blue-chip DeFi" is not infrastructure at all but a reflexive liquidity membrane that tears at the first correlated shock.

The attack itself lasted minutes. The aftermath is still reshaping how builders, auditors, and allocators think about cross-chain trust. And if LayerZero's preliminary attribution holds, the same North Korean unit that drained $285 million from Drift Protocol 18 days earlier just added another $292 million to its 2026 haul — bringing Lazarus's confirmed April take above $575 million through two structurally different attack vectors.

EigenLayer now secures more than $15 billion in restaked ETH across 40-plus registered Actively Validated Services. That is more capital than the national bank reserves of many small countries — mobilized, slashable, and theoretically working. But after three years of growth, one uncomfortable question is forcing itself to the surface: how much of this security is actually being paid for?

The answer, as of April 2026, is "less than you'd think." A small cluster of AVSs — led by EigenDA, and joined by the newer EigenAI and EigenCompute — generate real economic fees. The rest, by and large, pay operators with EIGEN emissions, points programs, and airdrop expectations. ELIP-12, the December 2025 governance proposal now rolling into effect, is the protocol's first serious attempt to separate the two camps. The reality check has arrived.

The $15B Number and What It Hides​

EigenLayer's headline TVL — $15.258 billion in restaked ETH, roughly 4.36 million ETH — looks like validation of the restaking thesis. ETH holders get a second yield on top of base staking; AVSs get pooled economic security without bootstrapping their own validator sets; Ethereum wins a new layer of credibly neutral infrastructure. Everybody in the flywheel gets paid.

The problem is the word "paid." Restaking yields come from two very different sources. The first is genuine AVS fee revenue — users of a service sending ETH, stablecoins, or AVS-native tokens to operators in exchange for the work done. The second is emissions — EIGEN token incentives, points, or treasury-funded rewards that AVSs use to attract operator stake before they have any customers.

From a restaker's wallet, the two look identical. From an economic-sustainability standpoint, they could not be more different.

Who's Actually Generating Fees​

Strip out emissions and the AVS revenue picture collapses dramatically. The fee-paying cohort in 2026 looks like this:

• EigenDA is the flagship. Mantle Network uses it as its primary data availability layer, with roughly $335 million in restaked assets backing Mantle's DA and a 200-plus operator set. Celo and a handful of other rollups pay EigenDA for throughput that clocks in at 15 MB/s versus Ethereum's native 0.0625 MB/s. This is real revenue, from real rollups, at volumes that grow as L2 activity grows.
• EigenAI went live on mainnet in late 2025, offering verifiable AI inference — an OpenAI-compatible API that guarantees prompts, models, and responses are unmodified and reproducible across runs. Early customers are paying for deterministic inference that centralized LLM providers structurally cannot offer.
• EigenCompute entered mainnet alpha in January 2026, handling off-chain execution verification. It is the newest revenue line, and the one most dependent on enterprise adoption to prove out.

Everything else — the long tail of 30-plus registered AVSs — earns little to no fee revenue. Their operators are paid primarily in EIGEN emissions, team-treasury rewards, or expectations of future value. This is not hidden; Eigen Foundation itself has acknowledged it by moving to restructure how emissions are distributed.

The Power Law Is the Story​

AVS revenue concentration in EigenLayer mirrors a pattern that plays out almost everywhere in crypto. Look at Ethereum Layer 2s: Base alone accounts for close to 70% of total L2 fee revenue, generating about $147,000 in daily fees versus Arbitrum's $39,000. Only three L2s clear $5,000 per day. The rest are rounding errors.

Polkadot's parachain model shows the same shape — shared security, a small cluster of parachains doing most of the economic work, a long tail of auction winners who never produced sustainable demand. Shared-security ecosystems appear to structurally concentrate around a few high-fee applications. EigenLayer is following the same curve.

Which forces a narrative question: if $15B in restaked ETH is available as security but only 3-5 AVSs generate real fees, is restaking creating genuine security infrastructure — or is it, functionally, a yield-generation mechanism for ETH holders who wanted staking alternatives and got them wrapped in a security narrative?

The most honest answer is "both, for now." EigenDA is genuine critical infrastructure for a growing set of rollups. EigenAI is solving a real problem for AI applications that need verifiable inference. Those services justify the restaking thesis. The long tail does not — yet. Whether it ever will depends on which way the incentives finally point.

ELIP-12: The First Hard Cut​

That is what the December 2025 ELIP-12 proposal is trying to fix. The core mechanics are blunt:

• A 20% fee on AVS rewards that are subsidized by EIGEN emissions, funneled into a fee contract designed for potential EIGEN buybacks.
• Only fee-paying AVSs remain eligible for staker and ecosystem incentives. If your service doesn't generate real fees, you don't get to subsidize operators with EIGEN from the treasury.
• 100% of EigenCloud service fees (EigenDA, EigenAI, EigenCompute), after operational costs, routed toward buybacks — tying token value directly to service revenue.
• A new Incentives Committee to set emissions policy, staffed by Eigen Foundation and Eigen Labs.

The design intent is explicit: emissions should reward AVSs that attract productive stake and generate real revenue, not AVSs that exist as security theater. The Eigen Foundation has stated that rewards "may be reduced to idle capital that does not secure AVSs."

Read another way: EigenLayer is instituting a minimum viable revenue threshold, in all but name. It is a concession that the "40-plus AVSs" number was always partly a vanity metric, and that the ecosystem's real value is concentrated in a smaller, harder core.

What a Mature Restaking Ecosystem Looks Like​

If ELIP-12 works as designed, the medium-term picture is a consolidation, not a collapse. Expect the AVS count to fall — some services will fail to generate fees and lose incentive eligibility, some will quietly unwind — while the surviving core gets meaningfully better resourced. The likely shape:

1. EigenDA keeps scaling throughput from today's 50 MB/s toward a targeted several hundred MB/s and sub-second latency, picking up additional rollup customers as the cost curve improves against Celestia and alternative DA layers.
2. EigenAI and EigenCompute grow as verifiable AI moves from crypto-native demand into enterprise AI pipelines that need deterministic inference and proof-bearing compute.
3. Vertical AVSs in specialized domains — oracle networks, cross-chain bridges, MEV infrastructure — survive if they attract paying users, and die if they don't, regardless of how much EIGEN they can afford to emit.
4. Restaking yields normalize downward as the share of yield that comes from genuine fees grows and the share from emissions shrinks. Yields will feel less punchy but be more durable.

The bear case is that fee revenue simply never grows fast enough to justify the $15B backing. In that world, ETH holders gradually rotate capital back to base staking or LSTs, restaking TVL shrinks, and EigenLayer consolidates as specialized infrastructure for DA and verifiable AI rather than as "the new trust marketplace for the internet." That is not a failure — it is just a smaller story than the initial pitch.

What Builders Should Take From This​

For teams deciding whether to launch as an AVS, the implications are sharpening fast:

• Budget for fee revenue from day one. EIGEN emissions are no longer a free growth lever; ELIP-12 gates them behind real fee generation. An AVS without a fee model is, going forward, an AVS without a future.
• Assume the tail compresses. If your thesis depends on staying a "registered AVS" with no users, recalibrate. The emissions committee will not fund pure optionality.
• Pick a vertical with measurable demand. Data availability, AI verification, and compute have paying customers today. Generalized "restake my ETH here for future security demand" narratives are on borrowed time.

For ETH holders and restakers, the cleaner question is whether the yield you are receiving is durable. If most of it comes from emissions of a specific AVS's native token, treat it as a time-limited subsidy and size accordingly. If it comes from EigenDA fees or EigenCloud service revenue, treat it as closer to real yield — still subject to protocol risk, but not structurally short-lived.

The restaking narrative in 2024 sold pooled security as a general-purpose primitive. The 2026 reality is more specific and, arguably, more honest: restaking is infrastructure for a small set of services that can actually pay for security. That is a smaller claim than "the marketplace for decentralized trust," but it is one the numbers will actually support.

BlockEden.xyz runs reliable Ethereum and L2 RPC infrastructure for teams building on top of the restaking and rollup stack. Explore our API marketplace to ship production services backed by an infrastructure partner that cares about the same sustainability questions you do.

Sources​

• EigenLayer's TVL crosses $15 billion as restaking protocol expands ecosystem — The Block
• EigenLayer expands restaking links with Mantle and ZKsync — Blockworks
• Celestia vs EigenDA: Choosing the Best Modular DA Layer — DA Layers
• EigenLayer's Double Launch: 4x Rewards and Verifiable AI Infrastructure — P2P.org
• EigenLayer Proposes ELIP-12 To Reshape EIGEN Incentives — OurCryptoTalk
• Foundation behind restaking protocol EigenLayer plans bigger rewards for active users — CoinDesk
• Base Leads L2 Fees With $147K Daily as Most Chains Earn Under $5K — Bitcoin Ethereum News
• EigenCloud Latest Updates — CoinMarketCap