Stratum V2 in 2026: Inside Bitcoin Mining’s Protocol Rewrite

Every ten minutes or so, one machine somewhere on Earth wins the right to write Bitcoin’s next block. The machine that wins almost never decides what goes inside it. Which transactions are included, and in what order, is chosen by whatever pool the miner has pointed its hashrate at, not by the miner itself. Stratum, the protocol that connects a mining rig to a pool, has quietly made that the default arrangement since 2012. Stratum V2 is the ground-up rewrite that tries to change who holds the pen.

The stakes are easy to price. Bitcoin trades near $85,970 as this is written, a market value close to $1.73 trillion, and the network runs at roughly 945 EH/s. The block subsidy of 3.125 BTC plus fees means miners share about 450 BTC of fresh issuance every day, close to $38.7 million. That reward flows through a leaderboard where the top three pools, Foundry USA, AntPool, and F2Pool, together direct just over 61% of the hashrate, and AntPool has climbed to 21.2%, narrowing the gap on Foundry. The concentration is well documented. What gets far less attention is the plumbing beneath it: the actual wire protocol that decides who assembles the block. This piece is about that wire, and about the rewrite that is slowly replacing it.

HOGE Wire has covered pool concentration and the 51% question before. This is the layer underneath: the bytes on the connection between a miner and its pool, and a narrower question than who controls the hashrate. What does Stratum V2 actually change, how does it work, and how close is it to mattering in practice?

What a Mining Protocol Is Actually For

A mining pool is a coordination layer. It takes thousands, sometimes millions, of ASIC chips owned by many different people and points them at the same goal, then splits the winnings smoothly so no single miner has to wait years for a payout. To do that, the pool and every machine have to talk constantly. The pool hands out work: a block header template and a slice of the nonce space to search. Each machine grinds through hashes, and when one clears the pool’s share threshold it reports back. The pool tallies shares, and when a machine finds a hash that satisfies the real network difficulty, the pool has found a block and pays everyone in proportion to the work they contributed.

The protocol carrying that conversation is Stratum. It was introduced in 2012 by Marek Palatinus, known as Slush, for the pool he had launched in 2010 (Slush Pool, rebranded Braiins in 2022, the first mining pool Bitcoin ever had). Stratum replaced an older, wasteful polling method with a push model: the pool notifies miners of new work the instant it has it, and miners submit shares as they find them. It is plain JSON-RPC over a TCP socket, which is a large part of why it won. It is trivial to implement, easy to debug, and readable by a human. For more than a decade it has been the near-universal language of industrial Bitcoin mining.

Simplicity was the right trade in 2012, when a pool was a hobbyist project and the whole network hashed at a rate a single modern rig would embarrass. It is a worse trade now, with tens of millions of dollars in daily issuance riding on the wire and hashrate concentrated in a handful of operators. Stratum V1 baked in three assumptions that were harmless then and expensive today.

The Three Things Stratum V1 Got Wrong

The first problem is that Stratum V1 is plaintext. It has no encryption, no authentication, and no integrity checking. Every message between a miner and its pool travels in the clear, which means anyone positioned on the network path can read it and, worse, rewrite it.

The second problem is that the pool builds the block. Under V1, the template, the actual set of transactions that will go into the next block, is assembled by the pool and handed down to the miner as a finished job. The miner supplies the electricity and the silicon; the operator supplies the transaction selection. The miner is grinding nonces on someone else’s block.

The third problem is that V1 is chatty. JSON-RPC is verbose, human-readable text, and at fleet scale the bandwidth and the latency of shipping all that text add up. Every millisecond a miner spends working on stale work, a job for a block that has already been found, is wasted electricity.

Stratum V2 is a from-scratch answer to all three, and it is worth taking them one at a time, because the encryption fix, the efficiency fix, and the block-building fix are technically independent and are being adopted at very different speeds.

Hashrate Hijacking: Why Plaintext Mining Is a Liability

Start with the security hole, because it is the least discussed and the most concrete. Since Stratum V1 traffic is unencrypted and unauthenticated, a party that can sit between a miner and its pool (a compromised router, a malicious or coerced ISP, a hijacked BGP route, a rogue employee inside a hosting facility) can quietly interfere with the connection. The classic version of the attack redirects a slice of the victim’s submitted shares to the attacker’s own worker, or rewrites the payout address, siphoning a fraction of the miner’s output without ever taking the machine offline.

Spark’s protocol research describes hashrate hijacking on V1 as an active attack vector, not a theoretical one, with typical losses of up to 2% of a facility’s hashrate before the theft is detected. Two percent sounds ignorable until it is priced. Skimming is designed to hide inside the natural variance of pooled mining, where daily luck swings far more than 2% anyway, so a slow leak can run for a long time before an operator distinguishes it from bad luck. On a large facility that is real money walking out the door disguised as noise.

Stratum V2 removes the entire class of attack by encrypting and authenticating the channel from the first byte. There is no plaintext to intercept, no unauthenticated message to forge, and no way for a man in the middle to rewrite a job or a share without the tampering being detected. This is the part of V2 with no governance question attached: no change to who builds blocks, no philosophical argument about decentralization, just a strictly better wire. It is also, for exactly that reason, likely to be adopted first.

Inside the Binary Wire

Stratum V2 throws out JSON-RPC text and replaces it with a compact binary framing. Spark measures the binary encoding at roughly 70% smaller than the equivalent V1 JSON, which cuts bandwidth by around 60% on the pool side and 70% for miners, according to figures cited when the major pools announced their support. Smaller messages also travel faster: where V1 job propagation sits around 228 milliseconds, V2 is engineered to deliver a new job in under 2.5 milliseconds. That latency gap is not cosmetic. Faster job delivery means a miner spends less time hashing on stale work, and less stale work is the same thing as higher effective revenue; the pools that backed V2 put the combined efficiency gain at up to 7.4% higher profitability.

The channel is also encrypted and authenticated, and V2 does it with primitives Bitcoin already uses. The handshake runs on the Noise Protocol Framework with the cipher suite Noise_NX_Secp256k1+EllSwift_ChaChaPoly_SHA256. In plain terms: the key exchange uses secp256k1, the same elliptic curve that secures every Bitcoin transaction; the traffic is encrypted with ChaCha20-Poly1305 authenticated encryption carrying a 16-byte message authentication code; and hashing is SHA-256. The pool authenticates itself to the miner with a 64-byte Schnorr signature under BIP 340, the signature scheme Bitcoin adopted at the Taproot upgrade. The result is a connection where the miner is certain it is talking to its real pool, the pool is certain the shares are genuine, and nobody on the path can read or alter either side.

The Sub-Protocols: How Stratum V2 Is Wired Together

Stratum V2 is not one monolithic protocol but a small family of them, split cleanly by job. Understanding the pieces is the difference between treating V2 as a slogan and understanding what it actually rearranges.

• The Mining Protocol is the core. It carries jobs down to mining devices and share submissions back up, and it supports several channel types (standard, extended, and group channels) so a single connection can serve one ASIC or aggregate an entire farm.
• The Job Declaration Protocol is the headline. It lets a miner declare a block template it built itself to the pool, which then validates and permits it. This is the mechanism that moves transaction selection from the operator to the miner.
• The Template Distribution Protocol connects a miner’s own Bitcoin full node to the software that turns a raw template into a mineable job. The full node, playing the role of Template Provider, is where the transaction set actually comes from when a miner builds its own block.
• Job Distribution, handled by the Job Declarator role, coordinates the declared jobs between the miner’s side and the pool.

The roles map onto software you can actually run: a Template Provider (a full node, with modern Bitcoin Core builds able to serve templates directly), a Job Declarator (which declares those templates to the pool), the Pool (which validates and accounts for payouts), the Mining Device (the ASIC), and a Translator Proxy that lets legacy V1 hardware speak V2 upstream. The whole design has one organizing idea: transaction selection and payout accounting, fused together inside the pool under V1, are pulled apart into separate roles that can be run by different parties. That separation is the entire point.

Job Declaration: Handing the Pen Back to the Miner

Job Declaration is the feature that turns Stratum V2 from an efficiency upgrade into a structural change. With it, a miner runs its own full node, builds its own block template locally by selecting transactions straight from its own mempool, and declares that template to the pool. The pool’s role shrinks: it validates that the template is well-formed and that the coinbase output pays what the pool’s accounting expects, and then it lets the miner grind on that block. Crucially, the pool does not modify the transaction list. If the miner’s machine wins, the block that lands on the chain reflects the miner’s transaction choices, not the operator’s.

That is the inversion. For fourteen years, supplying hashrate and choosing the contents of a block have been bundled into a single act performed by the pool. Job Declaration cuts the bundle in half and gives the transaction-selection half back to the miner. Mark Zalan, CEO of GoMining, put the old arrangement bluntly after his firm built the template for the first production Job Declaration block: “For years, mining pools have determined which transactions are included in Bitcoin blocks.” The demonstration was the point, showing that a miner, not a pool, could decide what a block contains while still collecting smoothed pool payouts.

It is worth being precise about what this does and does not fix, because the two are easy to conflate. Job Declaration does not make hashrate less concentrated. Foundry, AntPool, and F2Pool will still command the same share of the network’s machines the day after every pool ships V2 as they do today. What it changes is whether that concentration also means concentrated control over the contents of blocks, which is a different and arguably more important question.

The Reference Implementation Nobody Owns

A protocol is only as real as its code, and Stratum V2’s code is a public commons. The Stratum Reference Implementation, or SRI, is the open-source, community-maintained implementation of the full V2 stack, written in Rust with a minimum supported compiler version of 1.75. It ships the low-level building blocks (the binary codec, the Noise encryption layer, message framing and parsing) and, in a companion applications repository, the higher-level roles a real deployment needs: a Pool, a Job Declarator, a Translator Proxy, a test Mining Device, and the hooks to a Template Provider.

The Translator Proxy deserves a callout, because it is the bridge that makes adoption plausible rather than utopian. It lets a fleet of stock V1 ASICs connect and speak V2 to an upstream pool without any firmware change on the machines themselves, which means the tens of exahashes of hardware already humming in warehouses can join a V2 pool without being ripped out and replaced. Some of the higher-level application components are still marked alpha, a candid reminder that the encrypted, miner-built-template future is under active construction rather than finished.

The working group behind the specification was founded in 2022 by Braiins, the company descended from the original Slush Pool, and Spiral, Block’s independent Bitcoin engineering arm. It has run since as an open-source effort rather than a vendor’s product, and the specification itself is published openly at stratumprotocol.org. No single company owns the protocol, which is precisely the property Bitcoin’s mining layer needs if the point is to reduce, not relocate, concentration.

From Spec to Blocks: the 2026 Timeline

For years Stratum V2 was a well-regarded specification that almost nobody ran in production. 2026 is the year that started to change, though not as fast as the headlines suggested.

The signal moment came in May 2026, when seven pools representing roughly 75% of global hashrate, Foundry USA, AntPool, F2Pool, SpiderPool, MARA Pool, Block Inc., and DMND, publicly joined the Stratum V2 Working Group. Andy Zhou, CEO of AntPool, said the pool was “proud to support the broader adoption of Stratum V2,” framing an open, interoperable standard as the way for the industry to collaborate on “efficiency, security, and decentralization.” A month later, on June 25, 2026, block 955,318 became the first known production Job Declaration block: the DMND pool mined it, and the template was built by the miner, GoMining, which used the block to include its own payment transactions. Alejandro De La Torre, CEO and co-founder of DMND, described it as “a miner just mined the first Stratum V2 block to power their own product end to end,” with “no pool in the way.”

Here is the part the announcements tended to blur: joining a working group is not the same as running the protocol. As of mid-2026 only two pools ran Stratum V2 in production with Job Declaration, and both are small: Braiins, the protocol’s originator, and DEMAND (DMND), the first V2-native pool, which opened to the public in November 2025. The giants that joined the working group in May, Foundry, AntPool, F2Pool, and MARA, were in testing and planning phases, not production. The gap between the 75% of hashrate that has endorsed V2 and the low-single-digit percentage that actually builds its own blocks is the real state of play, and closing it is the work of 2026 and 2027, not a switch that has already flipped.

The table below shows the live leaderboard on September 22, 2026, with each pool’s stance on miner-built templates. Shares and EH/s are from Hashrate Index on a network of about 945.7 EH/s.

DATUM: the Version-1-Compatible Shortcut

Stratum V2 is not the only route to miner-built templates. Ocean, the pool co-founded by longtime Bitcoin Core developer Luke Dashjr, took a different path with DATUM (Decentralized Alternative Templates for Universal Mining), and the contrast is instructive.

DATUM reaches the same destination, letting the miner choose the transactions, but it gets there by layering on top of the existing Stratum V1 protocol instead of replacing it. A miner runs a full node plus a DATUM gateway; the gateway builds the block template locally and feeds it into an ordinary V1 mining connection. The critical consequence is that DATUM needs no new ASIC firmware. It runs on the hardware that already exists, today, which is why Ocean was producing miner-built blocks via DATUM back in 2024, well before the first Stratum V2 Job Declaration block. Dashjr has framed the payoff in decentralization terms, writing in 2026 that with independent miners building their own templates, “instead of two blocks made by the same pool, we get blocks made by two independent miners, exactly the same as if they weren’t using any pool at all.”

The tradeoff is real and cuts the other way on security. DATUM inherits Stratum V1’s plaintext channel; it solves the block-building problem without solving the encryption problem. Stratum V2’s Job Declaration is the cleaner, more complete design (encrypted, authenticated, and miner-built), but it depends on firmware most ASICs do not yet run. DATUM trades architectural elegance for the ability to work now. In a network where the installed base is enormous and firmware upgrades are slow, that is not a small advantage.

Why Template Control Is Worth This Much Effort

It is fair to ask why so much engineering is being spent on who assembles the block, as opposed to who gets paid. The answer is that building the block is where value and power quietly concentrate.

Whoever assembles the template chooses which fee-paying transactions to include and in what order. That is precisely the surface where transaction-ordering value, the Bitcoin cousin of the MEV that reshaped Ethereum, is captured or ignored. As the block subsidy keeps halving every four years, a schedule whose economics HOGE Wire has examined in its coverage of the halving cycle, transaction fees become a larger slice of miner revenue, and the right to order transactions becomes proportionally more valuable. Under Stratum V1, that right belongs to the pool operator, not to the miner who supplied the hashrate and bears the electricity bill.

Concentration is what turns this from a fairness footnote into a systemic concern. With the top three pools directing just over 61% of hashrate and a Nakamoto coefficient of 3 (meaning as few as three operators, acting together, could command a majority of block space), template authority is concentrated in the same few hands as hashrate. Stratum V2 and DATUM do not dilute the hashrate concentration; what they do is sever the link between it and template authority, so that a concentrated payout layer no longer implies a concentrated block-building layer. That decoupling is the whole reason the protocol fight is worth having.

The Censorship Flashpoint

The most vivid illustration of why the pen matters is transaction censorship, which is not hypothetical. Because a Stratum V1 operator builds every block for the hashrate pointed at it, it can also filter transactions out of those blocks, and at least one large pool has. F2Pool, a top-three pool, was found filtering transactions associated with US sanctions lists, first documented by the researcher known as 0xB10C in late 2023. Co-founder Chun Wang acknowledged the compliance filter, then disabled it after community pushback, arguing that a censorship-resistant system must be designed to resist censorship at the protocol level, rather than relying on each participant to act conscientiously.

That argument is the entire case for Stratum V2 compressed into one line. If the operator builds the block, then neutrality depends on the operator’s goodwill, and goodwill is a thin defense against a subpoena. If the miner builds the block, a single pool can no longer quietly drop transactions for all the hashrate it coordinates, because the filtering decision is distributed across thousands of independent template builders instead of concentrated in a few operators. This is exactly the seam where sanctions-compliance pressure, a Treasury and OFAC matter that sits inside the broader enforcement machinery HOGE Wire has traced through tools like Section 311, meets Bitcoin’s claim to neutrality. Stratum V2 and DATUM answer a policy problem with a technical mechanism: make censorship structurally hard rather than merely unfashionable.

Does Building Your Own Block Change the Legal Picture?

For miners in the United States, the regulatory frame around all of this is unusually settled by crypto standards. In a statement dated March 20, 2025, the SEC’s Division of Corporation Finance said that proof-of-work mining, explicitly including both solo mining and participation in a mining pool, is not a securities transaction: the rewards come from the miner’s own computational contribution, an administrative or ministerial act, rather than from the entrepreneurial or managerial efforts of some third party, so it does not satisfy the Howey test. A broader interpretive release the following year, issued March 17, 2026, extended the no-securities view across mining, staking, and related activities, with SEC Chairman Paul Atkins framing it as drawing clear lines in clear terms, which he called what regulatory agencies are supposed to do.

Job Declaration does not disturb any of that. Choosing your own transactions and building your own template is, if anything, an even clearer example of your own computational effort rather than reliance on someone else’s enterprise. The live regulatory pressure on template control is not securities law at all. It is sanctions compliance, which runs through Treasury and OFAC rather than the SEC, and, for anyone trading the hashrate futures and other derivatives that let miners hedge, the CFTC. The protocol change is orthogonal to the Howey question; a miner who upgrades to Stratum V2 is in exactly the same securities posture as one who does not.

What Has to Happen for Stratum V2 to Win

The bottleneck for Stratum V2 is not the specification, which is mature, but firmware and defaults. Three things have to line up before V2 is the network’s normal case rather than its frontier.

First, ASIC firmware has to ship V2 by default. BraiinsOS+, LuxOS, and Vnish already support it natively; stock Bitmain firmware still needs the SRI Translation Proxy to bridge onto a V2 pool, which works but adds a moving part. Second, the large pools have to move from working-group membership to production Job Declaration, a step most had not taken as of mid-2026. Third, and most easily overlooked, enough miners have to actually run their own full node and build their own templates. Stratum V2 permits miner-built blocks; it does not force them. A miner can run a V2 connection and still accept pool-built jobs out of habit or convenience, in which case the encrypted channel arrives but the decentralization does not.

SRI’s own projection is that V2 becomes the default protocol for new ASIC firmware shipments by the end of 2026, potentially reaching 40% to 60% of network hashrate. Treat that as a target rather than a forecast; the same pools that endorsed the standard in May were, months later, still testing. The safe prediction is that the pieces arrive at different speeds. The encrypted, authenticated channel is a drop-in improvement with no governance question, and it will likely spread fastest. Miner-built templates ask miners to run infrastructure they have outsourced to pools for more than a decade, and that cultural change will lag the technical one. Bitcoin’s longer roadmap, from covenant proposals to a fee-driven security budget after the subsidy fades to projects like Babylon that lease Bitcoin’s security to other chains, all quietly assume a mining layer that stays neutral and hard to capture. Stratum V2 is the least glamorous piece of that assumption, and possibly the most load-bearing.

Frequently Asked Questions

What is Stratum V2 in simple terms?

Stratum V2 is a rewrite of the protocol that connects Bitcoin miners to mining pools. It encrypts the connection, shrinks the data sent over it, and, most importantly, lets a miner build its own block template instead of accepting whatever transactions the pool chooses. It is being adopted gradually, with the encryption arriving faster than the miner-built-template feature.

How is Stratum V2 different from Stratum V1?

Stratum V1, introduced in 2012, is plaintext JSON in which the pool assembles every block and the miner only grinds nonces. Stratum V2 is a compact, encrypted binary protocol that can hand transaction selection back to the miner through a feature called Job Declaration, while also cutting bandwidth and latency.

What is the difference between Stratum V2 and DATUM?

Both let miners build their own block templates. Stratum V2 Job Declaration is a full protocol rewrite that needs new firmware or a translation proxy, and it encrypts the channel. Ocean’s DATUM layers template building on top of the existing Stratum V1 protocol, so it works on today’s hardware without a firmware change but keeps V1’s unencrypted channel.

Does Stratum V2 fix Bitcoin mining centralization?

Not the hashrate concentration itself; the same pools still command the same share of machines. What Stratum V2 changes is that block construction is decoupled from the pool, so a concentrated payout layer no longer means a concentrated block-building layer, but only if miners actually run their own templates rather than accepting pool-built ones.

Is Bitcoin mining regulated by the SEC?

The SEC has stated that proof-of-work mining, whether solo or through a pool, is not a securities transaction. Sanctions compliance falls to the Treasury Department and OFAC, and hashrate derivatives fall to the CFTC, so the securities regulator is largely not the relevant authority for mining itself.

By Yuki Tanaka, mining and infrastructure desk, HOGE Wire.