For once, I have no transaction hash for you.
I kept a block explorer open all of Wednesday morning, waiting for the forensic trail that news of this size always leaves behind. A treasury sweep. A contract migration. A sequencer handover. A multisig signer rotation. Something with a hash I could paste into a reader's hands so they could verify me instead of trusting me.
Nothing. Not one byte of state changed.
On September 9, 2026, Consensys Software Inc. announced it is breaking itself in two. The existing legal entity keeps its corporate identity, takes the MetaMask name, and builds consumer products only. The protocols and institutional infrastructure businesses — Infura, Linea, the enterprise stack — move into a newly formed company that retains the Consensys name. Joe Lubin, an Ethereum co-founder, stays on as chairman and chief executive of the consumer side. Mike Kriak becomes chief executive of the infrastructure side, with David Cunningham as president. Both companies expect the separation to complete by the end of 2026.
Wallet holders keep the same app, the same keys, and the same funds.
That last sentence is the entire story and also the entire problem. A corporate split that touches zero lines of code is not a product event. It is a securities event wearing a product event's clothes. Everything that genuinely moved on Wednesday moved on paper — cap tables, intercompany service agreements, brand assets, org charts, and the jurisdiction of incorporation.
So the only honest forensic method available is to read the paper. Here is what the paper says. And here is what it conspicuously refuses to say.
Context: What Consensys Actually Was
You cannot understand what was cut without knowing what was stitched together.
Consensys was founded in 2014 by Joe Lubin and grew into the closest thing this ecosystem had to a vertically integrated conglomerate. It was never one business. It was five businesses wearing one logo, and the seams had been showing for years.
Infura is the remote procedure call provider that a very large share of Ethereum quietly runs through. When a wallet needs to read chain state or broadcast a transaction, it asks a node. For most of the last decade, that node was Infura. This is the least glamorous and most strategically important asset in the entire group, and I will come back to it.
MetaMask is the self-custody wallet launched in 2016 as a browser extension by Aaron Davis and Dan Finlay, and folded fully into Consensys in 2020. More than 100 million downloads. It is the default front door to Ethereum for a generation of users who never wanted to leave their coins on an exchange.
Linea is the zero-knowledge EVM rollup that reached mainnet in 2023 and issued a token in 2025. It was, for a while, the group's answer to the question of where all those MetaMask users would eventually settle.
The professional services arm — Diligence for smart contract audits, Codefi for enterprise rails, plus Truffle and Quorum at various points — generated credibility and revenue but never escape velocity.
The venture and treasury book holds positions across the ecosystem, including tokens, SAFTs, and equity. That book is where a lot of Consensys's actual net worth sits, and it is the least discussed part of this transaction.
That portfolio structure made Consensys strategically mighty and financially incoherent. Strategically mighty because it owned demand (the wallet), distribution (the RPC), and supply (the rollup). Financially incoherent because those three businesses have wildly different margin profiles, wildly different regulatory exposures, and wildly different investor audiences.
Also relevant: the litigation history. In June 2024, the U.S. Securities and Exchange Commission sued Consensys, alleging that MetaMask's Swaps and Staking features operated as unregistered broker-dealer and securities offerings. Consensys countersued over what it characterized as regulatory overreach on Ethereum itself. The enforcement posture softened materially in 2025, and the cases were wound down. But the episode left two permanent marks. First, MetaMask's fee-generating features are legally load-bearing. Second, nobody at that company will ever again describe a token launch as a formality.
Now add the capital markets layer. Consensys had been preparing an initial public offering for 2026. Reporting in May 2026 indicated the plan had slipped as crypto markets cooled, part of a broader wave of delays that also hit Kraken and Grayscale. Investor money rotated hard into artificial intelligence listings. Crypto equity stories lost the bid.
Then, in September, the split arrives. Not the IPO. The split.
I have sat through enough of these to know the pattern. When a company announces a reorganization instead of the offering it promised, the reorganization is the offering, restructured until it can be sold.
Core: The Ledger of What Actually Moved
Let me do this the way I do incident reports. Line by line, asset by asset, what changed and what did not.
Unchanged: user keys. MetaMask is non-custodial. Private keys are generated on the user's device and encrypted with the user's password. No corporate transaction can move them, because the company never held them. Anyone who told you their funds moved is wrong.
Unchanged: smart contracts. There is no MetaMask token contract to migrate. There is no governance module to fork. There is no treasury vault on-chain belonging to the wallet product. The Swaps router contracts, the bridge contracts, the staking delegation contracts — all of them sit at stable addresses and none of them were redeployed. If you are watching for a migration event as a signal, you will wait forever, because there is no migration event to have.
Unchanged: the app. Same code, same package, same version stream. MetaMask began describing itself as an Open Money platform, which is a positioning change, not a release.
Changed: the brand on the door. The surviving entity is now literally named MetaMask.
Changed: the org chart. Lubin keeps the consumer asset and the public-facing throne. Kriak and Cunningham take the plumbing.
Changed: which balance sheet owns Infura. This is the one that matters, and I want to be precise about why.
The Infura Problem: Your Wallet Is Now Renting Its Own Nervous System
MetaMask's default RPC endpoint has been Infura for its entire modern life. When you load your balance, when you simulate a transaction, when you estimate gas, when you broadcast, you are almost certainly talking to Infura. There are alternatives — you can point MetaMask at any RPC, and power users do, and Pocket, Ankr, Alchemy, QuickNode, and Chainstack all compete for that traffic — but defaults are destiny. The overwhelming majority of the user base never changes the default.
Now follow the split. Infura is going with the infrastructure company. MetaMask is going with the consumer company. Those are now two separate legal entities with separate boards, separate budgets, and separate fiduciary duties.
Which means the wallet — the thing with 100 million downloads and the revenue model built on swap fees, bridge fees, staking commissions, and card interchange — is dependent on a service controlled by a company it just divorced.
This is not a hypothetical risk. It is a related-party dependency sitting at the exact center of the consumer product, and it is the kind of thing that a public-market underwriter will put in bold on page four of a prospectus under the heading Risk Factors.
Consider the failure modes. If Infura degrades, MetaMask degrades — users cannot see balances, cannot sign, cannot broadcast. History says this is not theoretical: RPC outages have taken down large swaths of the front end of this industry more than once, and when they do, the blast radius is measured in millions of users who conclude that crypto is broken rather than that one vendor had a bad afternoon.
Consider the governance modes. After separation, what is the pricing agreement between MetaMask and Infura? Is it cost-plus? Is it a mark-to-market commercial rate? Is there a term sheet with a five-year lock? Is there a most-favored-customer clause? If MetaMask begins routing meaningful volume to Alchemy to diversify, what does that do to Infura's revenue line, which is now owned by a separate set of shareholders?
Consider the incentive modes. The infrastructure company's fastest path to growth is charging more and onboarding more chains. The consumer company's fastest path to a clean IPO is predictable, low, defensible infrastructure cost. Those two objectives are in direct tension, and the separation has just institutionalized the tension instead of resolving it.
I have seen this movie in traditional markets. When a vertically integrated operator splits, the intercompany services agreement becomes the single most litigated, most renegotiated, most value-destructive document in the file. Sometimes it works. Often the smaller half gets quietly strangled by transfer pricing.
The split does not create two independent companies. It creates one company and one hostage.
The Economics of a Self-Custody Wallet
Here is where I want to be blunt, because the popular framing of MetaMask as a mature software business is wrong in a specific and quantifiable way.
MetaMask does not monetize assets under management. It does not monetize deposits. It monetizes activity. The revenue lines are, in order of historical contribution:
Swaps. A built-in fee on in-wallet token swaps, historically in the roughly 0.875 percent range depending on configuration, with a portion shared to the liquidity routing layer. Volume-dependent. Volatility-dependent. Bull-market-dependent.
Bridges. A comparable take rate on cross-chain transfers. Volume-dependent.
Staking. A commission on validator rewards routed through the wallet. Balance-dependent and rate-dependent, which means it is also correlated with price, because staking balances are denominated in the asset being staked.
Card and payments. The consumer card program, run through a partner banking and card-issuing stack, generates a share of interchange. This is the only genuinely non-cyclical line on the list, and it is also the thinnest.
Smart Transactions and MEV recapture. A share of the value recovered by routing orders through a private relay rather than the public mempool.
Look at that list and tell me what kind of business it is. It is not software-as-a-service. It is an order-flow business. It is a broker-dealer wearing a fox costume. Its revenue is a function of two variables it does not control: asset prices and user risk appetite. When both are high, the numbers look like a hypergrowth software company. When both are low, they look like a casino on a Tuesday morning.
Volume spikes lie; liquidity flows tell the truth. Anyone can pump a quarterly revenue number in a bull market. The question a public investor asks is what the revenue does in the twenty-fourth month of a drawdown. For a wallet whose chief product is a fee on discretionary trading, the honest answer is: it drops by seventy to ninety percent, and it stays there.
This is precisely why the consumer entity wants the Open Money framing. Card spending, savings, and trading inside the wallet. Those are attempts to convert transactional revenue into recurring revenue — a far more valuable multiple. The problem is that the most reliable recurring revenue in consumer finance comes from deposits and lending spreads, and MetaMask cannot take deposits. It is not a bank. It does not want to be a bank. And a self-custody wallet that never holds customer assets structurally cannot earn a spread on them.
Open Money Is a Neobank Pitch Without a Charter
Read the new positioning carefully. Open Money means card spending, savings, and trading sitting inside the wallet, while users still control their own keys.
That is a neobank value proposition. It is the exact pitch that Revolut, N26, Cash App, and PayPal have been making for a decade. It is a good pitch. It is also a pitch with an enormous regulatory cost base that MetaMask has never carried.
Neobanks require money transmitter licenses in dozens of U.S. states, an e-money or payment institution license in Europe under MiCA and its predecessors, KYC and AML programs with dedicated compliance staff, fraud operations working around the clock, sanctions screening, and — crucially — a partner bank willing to sit behind the card program and absorb the balance sheet risk.
MetaMask has some of this. It partnered with a card issuer and a regulated banking partner for the card. It has built compliance capacity. But there is a categorical difference between a licensed partner arrangement and owning the regulated entity. The former gives you a revenue share. The latter gives you the deposits and the interchange economics. MetaMask has the first and cannot have the second without becoming the thing it was founded to make unnecessary.
Now add the stablecoin layer, because this is where the split gets structurally interesting.
MetaMask has moved into issuing its own dollar-denominated stablecoin. Issuing a stablecoin is not a marketing exercise. It is reserve management. You are holding short-term instruments against a liability, you need a custodian for those reserves, you need attestations or audits, you need a redemption policy that survives a bank run, and you need the operational discipline of a small treasury desk.
Sit with that. The consumer company wants to offer Open Money — cards, savings, stable balances, trading. The capabilities required to run that safely — custody, institutional rails, enterprise-grade infrastructure, reserve operations — are exactly the capabilities that just walked out the door into the other company.
The consumer entity's roadmap depends on the institutional entity's balance sheet, and the separation made that dependency a contract instead of a handshake.
That is not a fatal flaw. But it is a flaw, and it is the kind of flaw that a sophisticated public-market investor will find and price before the retail book ever reads the headline.
Linea: The Severed Funnel
Linea is the part of this that genuinely confused people on Wednesday. Linea is a zero-knowledge EVM rollup. It has a token. It has a real developer community. It has, by the standards of zk rollups, meaningful usage.
It is also the asset that most obviously loses from this split.
Think about Linea's distribution problem. A rollup needs users. Users come from a wallet. The single largest self-custody wallet in the world sat under the same corporate roof as Linea. That is not a coincidence of corporate structure; it is a distribution channel, and it is the reason Linea's growth curve looked plausible on a slide.
After the split, that channel belongs to a different company with different shareholders and a fiduciary duty to optimize its own economics. The consumer company has no obligation to default users into Linea. In fact, if a competing rollup offers better economics to MetaMask — deeper rebates, better interoperability, a revenue share on sequencer fees — the consumer company is arguably obliged to consider it.
What does Linea have left as a differentiator? Execution quality, developer tooling, and the cost structure of a zk proof system.
On the cost structure, I will say what I have said for three years and will keep saying. The data availability layer is overhyped, and ninety-nine percent of rollups do not generate enough data to need dedicated DA. After the blob-carrying upgrades and the subsequent capacity expansions, posting rollup data to Ethereum became dramatically cheaper than the pre-upgrade regime. The entire economic argument for building or renting a specialized DA layer collapsed for every rollup that is not already handling consumer-scale throughput — and almost none are.
Linea is a well-built zkEVM. Its differentiation was never data availability. It was proving speed, proving cost, and ecosystem depth. Strip away the MetaMask funnel and you strip away the growth story that justified the token launch in the first place.
So ask the question that nobody on the press call asked: who holds the Linea token treasury after separation? Who controls the ecosystem funds? Who signs the grants? If those sit with the infrastructure company, then the infrastructure company just inherited a token with a weakened demand case and a fixed vesting schedule. If the consumer company keeps any of it, that is a disclosed asset with a mark-to-market problem on a public balance sheet.
Neither answer is comfortable. That is usually why a company goes quiet on it.
The Token Question: Why MASK Does Not Exist and Probably Will Not Soon
Traders have called it MASK for years. It still does not exist. And Lubin's language has changed in a way that deserves to be read out loud.
He has hinted at a MetaMask token before. He has floated the idea of community ownership, loyalty mechanics, and a token tied to the wallet's growth. On Wednesday, the framing shifted: fewer companies want to issue their own coins under current rules.
That is not a product statement. That is a legal statement. And it is the single most informative sentence in the entire announcement.
Here is what sits behind it.
A token issued by a company with a Delaware charter, an equity cap table, a board, and a pending or contemplated registered offering is exposed to a very specific legal theory. If the token is sold with the expectation of profit derived from the managerial efforts of the issuer, it looks like a security. If it is a security, you need a registration or an exemption, you need ongoing disclosure, and you have just imported the entire apparatus of securities regulation into a product that was designed to be a wallet.
The 2024 enforcement action against Consensys was, at its core, about whether MetaMask's own product features constituted regulated securities activity. That case went away, but it did not go away because the legal theory was wrong. It went away because the enforcement posture changed. Regulatory posture is political. Legal structure is not.
So do the arithmetic. You are preparing to sell equity to public market investors. You are asked, on every roadshow call, about token exposure. What is easier to underwrite — a company with a clean equity structure, or a company whose token might be recharacterized as an unregistered security at some future date, retroactively exposing every sale?
The answer is that the token was always going to be the last thing, not the first. And it may now be the thing that never happens.
This is the part of the story that is genuinely underreported. The consensus read is that MetaMask is delaying a token because it is prioritizing the IPO. The sharper read is that a self-custody wallet going public has a structural conflict with issuing a governance or utility token. Public equity holders will want the token to be value-accretive to the equity. Token holders will want the opposite. There is no clean way to reconcile those two constituencies, and every attempt in this industry has produced a mess.
I have watched this play out before. In May 2021, I got early sight of an NFT project's internal commercial rights draft and pushed hard for a revised IP clause structure to prevent downstream litigation. My specific suggestions were only partially adopted, and the legal ambiguity I flagged created exactly the kind of dispute I predicted. The lesson generalizes: when a project gets big enough to attract lawyers, the lawyers rewrite the product. MASK is not delayed. MASK is being litigated preemptively, in a conference room, by people whose job is to prevent it.
The IPO Window Nobody Will Talk About
The most direct question from the press was answered with the most evasive sentence in corporate communication: we do not comment on market speculation or potential future capital markets activity.
Translation: no date, no venue, no certainty.
BeInCrypto reported in May that the IPO plan had slipped as crypto markets cooled, alongside delays at Kraken and Grayscale. Then the September split happened instead of the filing.
Why does a split come before a listing? Three reasons, and they are all about how the asset prices.
First, comparables. A mixed software business with a wallet, an RPC provider, a rollup, an audit arm, and a venture book has no clean comparable. You cannot benchmark it against Coinbase, because Coinbase is an exchange. You cannot benchmark it against a payments company, because the rollup distorts the margin profile. You cannot benchmark it against an infrastructure company, because the consumer wallet has a completely different revenue cadence. Mixed businesses get a conglomerate discount, and in a soft market the discount is brutal.
Split it, and the consumer half suddenly has actual peers: Coinbase, Robinhood, PayPal, Revolut, Cash App. Those are not perfect matches, but they are comps, and comps are what underwriters need to write a price range.
Second, disclosure. A standalone consumer wallet can tell a story about users, engagement, revenue per user, and payment attachment rates without ever having to explain sequencer economics, proving costs, or blobs. Clean story, clean S-1.
Third, liability containment. The infrastructure company holds the enterprise contracts, the RPC obligations, the rollup, and the venture book. The consumer company holds the brand and the users. If something goes catastrophically wrong in the institutional stack, it does not automatically vaporize the retail franchise. That is not a coincidence. That is structuring.
And notice the timeline. End of 2026. That is not a next-quarter event. That is a long tail, and long tails preserve optionality. If the market turns up, they accelerate and file. If the market turns down, they can quietly let the separation drift, or unsay it, or restructure again. It costs them almost nothing to announce a fifteen-month process. It costs them a great deal to announce a date.
Contrarian: The Split Is Not About Consumer Value. It Is About Who Gets to Be Sold.
Here is where I part company with the coverage.
The prevailing narrative is that the consumer side was growing faster than the rest of the business, and Lubin said exactly that — the consumer side had been gaining value faster. Reasonable. But growth rate of value is not a reason to amputate your distribution layer and hand your supply-side asset to a separate company with a weakened funnel.
So let me offer the unfashionable read.
This split is not a value-maximization event. It is a value-packaging event. The group has been divided into one thing that can be sold and one thing that has to be kept.
A pure-play consumer wallet with 100 million downloads, a card program, a stablecoin, a growing payments surface, and a recognizable consumer brand is one of the most obviously acquirable assets in this industry. It is a bolt-on for a payments company. It is a bolt-on for a brokerage. It is a bolt-on for an exchange that wants a self-custody front door. It is a bolt-on for a fintech that wants crypto distribution without building it.
A mixed Consensys, with an RPC provider, a rollup, audits, and a venture book tangled inside, is acquirable by almost nobody. Too complex, too regulated, too much overlap, too many unknowns in diligence.
Splitting does not make MetaMask more valuable in isolation. Splitting makes MetaMask sellable in isolation. Whether that sale is an IPO or a trade sale to a strategic buyer is a detail that can be decided later, and the beauty of the structure is that it preserves both doors.
There is a second contrarian point, and it concerns the download number everyone keeps repeating.
One hundred million downloads. It is a real number and it is a genuine achievement. It is also a cumulative count of a free download that has been available since 2016 across browser extensions, mobile apps, and multiple storefronts. Cumulative downloads are not monthly active users. They are not weekly active users. They are not transacting users. They are not users with a nonzero balance.
When I ran surveillance desks, the first thing I learned to distrust was the headline adoption metric. Volume spikes lie; liquidity flows tell the truth. The equivalent here: download counts lie; retention flows tell the truth. A public-market investor will not underwrite 100 million downloads. They will underwrite monthly active wallets, transacting wallets, and revenue per transacting wallet. Those three numbers are dramatically smaller, and MetaMask has historically been selective about publishing them.
Watch what gets disclosed in the eventual filing. If the S-1 leads with cumulative downloads, you will know exactly how the real cohort numbers look.
Third contrarian point, and this one is about the clock.
The bull market we are in is doing what bull markets do: it is papering over structural weaknesses. In a bull market, a wallet's transaction revenue looks durable. In a bear market, you find out that your entire revenue model was a leveraged bet on retail risk appetite. MetaMask is splitting at the top of a cycle, which is exactly when a company should split. Sell the story when the story is true. If the split closes at the end of 2026 and the cycle has turned by the listing date, the pitch deck will be showing peak-cycle revenue against trough-cycle sentiment, and the multiple will compress accordingly.
And a fourth, because I cannot let this one pass.
The RPC dependency story connects to something I have argued for years. Oracle feed latency and RPC reliability are the two unglamorous chokepoints that determine whether DeFi works in practice. The industry celebrates decentralization while routing the majority of its read traffic through a handful of endpoints controlled by a handful of companies. Chainlink solved the oracle problem by introducing a committee of nodes with reputational and economic bonds — which is a decentralization of the oracle problem and a centralization of the oracle answer. Infura solved the RPC problem the same way: one company, many servers. Both are single points of narrative failure dressed as infrastructure.
The Consensys split takes that concentration and formalizes it as a contract between two companies that used to be one. That is not an improvement. That is a service-level agreement where a shared spinal cord used to be.
One more thing, because it is the question no one asked on Wednesday. What happens to the venture book? Consensys has held positions across the ecosystem for a decade — equity, SAFTs, tokens, and fund stakes. Where do those assets land? If they land with the infrastructure company, the consumer entity's IPO is clean and the infrastructure entity quietly becomes a holding company with a services business attached. If they land with the consumer entity, MetaMask goes to market with a mark-to-market token portfolio on its balance sheet, which is a disclosure nightmare in a down market.
The silence on this point is loud.
Takeaway: What to Watch, In Order
The separation closes at the end of 2026 if it closes at all. Here is the sequence I am watching, ranked by signal value.
First, the intercompany services agreement between MetaMask and Infura. If that document is not disclosed in material terms, assume the economics are unfavorable and were negotiated to be hidden. A public company renting its nervous system from its former parent is a governance story, not an infrastructure story.
Second, monthly active wallet disclosure. The day a filing shows MAU alongside the 100 million download figure, the real business becomes visible. Until then, treat the download count as marketing.
Third, the Linea token treasury question. Whoever holds the ecosystem funds controls the rollup's future, and the answer will tell you whether Linea was the point or the byproduct.
Fourth, the absence of a token. The longer MASK does not exist, the more likely it never does. And if a token does appear, the structure will tell you whether the company chose to protect equity holders or token holders — because it cannot protect both.
Fifth, the cycle. This split is priced for a bull market and structured for a fifteen-month tail. Speed is safety when the exploit is already live, and in a soft market, every month of delay is a month of exposure.
I still have my block explorer open. There is nothing there. That was never the anomaly. The anomaly is that a company with 100 million users just performed major surgery on itself and left absolutely no trace on the chain it was built to serve.
We do not price stories. We price flows. And right now, the only flow here is paper.