Hook
Consensys Software Inc. kept its legal identity and took a new name. The brand left the building. The protocols moved into a company that inherited the old one.
Read that sequence slowly, because it is the only structural fact in this week's announcement that carries legal weight. Consensys Software Inc. — same registered entity, same liabilities, same cap table — now answers to the name MetaMask. A newly formed firm takes the Consensys name, absorbs Linea, the institutional stack, and the developer tooling, and reports to a different chief executive. Mike Kriak runs it. David Cunningham sits as president.
No chain forked. No contract redeployed. Wallet holders keep the same app, the same keys, the same funds. The split is a re-labeling of legal entities precise enough to read like a proxy upgrade — the logic moves, the storage slot keeps its address.
Truth hides in the assembly, not the press release.
Context
MetaMask is the app more than 100 million people have downloaded to hold crypto without leaving coins on an exchange. Joseph Lubin, an Ethereum co-founder, has chaired the parent since its founding in the mid-2010s, when the company was positioned as the enterprise and developer arm of the Ethereum ecosystem. Over a decade that suite grew sideways: MetaMask for consumers, Infura for RPC access, Linea for cheaper execution on top of Ethereum, plus institutional and protocol work that never fit comfortably beside a wallet.
On Wednesday the fit stopped being a problem. The consumer platform becomes the whole company and markets itself as an "Open Money" platform — card spending, savings, and trading inside the wallet, keys still in the user's hands. Everything banks and developers buy moves out. Linea goes with it. Lubin stays on as chairman and chief executive of the MetaMask entity and cited the consumer side gaining value faster than the rest of the business. Both companies expect the separation to close by the end of 2026.
Two questions went unanswered, and they are the two that matter to anyone holding a term sheet.
The first is a listing. Consensys had been preparing an IPO in 2026. BeInCrypto reported in May that the plan had slipped as crypto markets cooled, part of a broader wave of delays that caught Kraken and Grayscale as well. Lubin declined to give a new date. A spokesperson went further: the company does not comment on market speculation or potential future capital markets activity.
The second is a token. Traders have called it MASK for years. It does not exist. Lubin had hinted at one previously. His language now is narrower — fewer companies want to issue their own coins under current rules.
Capital rotated into artificial intelligence listings this year. Crypto IPO calendars stalled behind it. A standalone consumer wallet may pitch more cleanly to public investors than a mixed software business. Whether it lists, and whether a token ever arrives, remains open.
Core
The first thing to test is the framing that this is a separation at all. Corporate shells do not separate; they reallocate. What reallocated here is the risk.
Beauty is the most sophisticated rug pull. The announcement is elegant — clean brand logic, a named leadership team, a stated completion window, and a reassurance that user funds never move. That elegance is doing real work. It is directing attention toward the consumer story and away from the cost curve that just changed hands.
Start with what MetaMask actually is as a business. It is a distribution surface that meters flow in basis points: swap fees on routed trades, on-ramp and off-ramp spreads, staking commissions, and now card interchange through the Open Money stack. Public equity markets understand basis points on consumer flow. They have priced that business model for fifty years. They do not understand sequencer revenue, proving systems, or RPC availability SLAs, and they discount all three heavily when they appear on the same income statement as a consumer app.
That is the re-rating. A pure-play consumer wallet gets a multiple. A mixed software company carrying zk infrastructure and institutional integration work gets a discount, because every analyst building a model has to write down the segment they cannot benchmark. Splitting does not create value. It stops one segment from suppressing the valuation of another. Lubin's stated rationale — the consumer side gaining value faster — is not the cause of the decision. It is the justification for a move that was already determined by how public markets price complexity.
Now the number that got repeated: more than 100 million downloads. I want to be precise about what that metric is. It is an installation count. It is not monthly active users. It is not transactions per wallet per month. It is not revenue per install. A download is a marketing event. A prospectus is built on retention, and the gap between the two is where most crypto listings of the last cycle went to die. Anyone modeling this entity will ask for cohort retention on a 30-day, 90-day, and 365-day basis. That data exists internally. It has not been published.
Then there is Linea, and this is where the split stops being a branding exercise.
Linea is a rollup. Rollups post data to Ethereum, and since EIP-4844 that data lives in blobs — a fixed number of them per block, priced through a separate fee market that mimics EIP-1559's burn-and-target mechanics. Blob supply does not scale with demand inside a given fork. It scales only when the protocol raises the target, which requires a hard fork and a consensus decision.
Demand is not waiting. Base, Optimism, Arbitrum, Linea, Scroll, zkSync, and a long tail of validiums using alternative data availability layers all compete for the same finite blob space. Every one of those teams is optimizing to post more data, more often, more cheaply. That is what rollups are for. It is also what guarantees the curve bends. My working position is that blob space saturates within two years, and when it does, rollup gas fees roughly double — not because anything broke, but because the auction clears at a higher price and every consumer of that fee market inherits the increase.
For the years after Dencun, blob fees sat at or near the floor and cheap execution looked structural. It was promotional. Linea's unit economics ride on that floor, and the floor is an artifact of temporary overcapacity, not a property of the design.
I have watched this movie before. In 2020 I spent two weeks inside a proposed Compound Finance governance upgrade and found an integer overflow that could have drained roughly $50 million. Nothing in the public documentation suggested risk. The vulnerability lived in the arithmetic of an edge case nobody had stressed. I reported it privately, they patched in 48 hours, and the community never learned the upgrade had been lethal. That is how upgrade risk usually behaves: invisible until the block where it is not.
Blob saturation is the same category of exposure at a different layer. It is not a bug. It is a scheduled cost increase that has no announcement date. Putting Linea into the protocols entity does not remove that cost. It removes it from the consumer story, which is a different and much smarter thing.
Now the part the announcement did not address at all: the dependency graph did not split with the entities.
MetaMask ships with a default RPC endpoint. Any user who has never manually changed a network configuration in the advanced settings is relying on infrastructure that now sits on the other side of a corporate boundary. The wallet's swap routing, its fiat ramps, its staking provider integrations, and its new card issuer all sit downstream of third parties — some of which are now "unrelated." A standalone consumer company leaning on its former parent for the single most latency-sensitive dependency in the product is a related-party arrangement, whether or not anyone calls it that. Arm's-length pricing on that contract is the first line item a public-market analyst will interrogate, and the second is what happens to the wallet if the counterparty reprices it.
There is a second conflict, and it is the reason the token question stayed unanswered.
An IPO and a token are not complementary instruments. They are competing claims on the same cash flow. MetaMask has debated a fee switch for years — whether to take a cut of the swap revenue it already routes. Enable it and the wallet has earnings. Those earnings belong to equity holders in a listed company. A tradable token that also captures value from that flow creates two classes of claimant on one revenue stream, and in most major jurisdictions the second one is a security with a disclosure obligation the first one already satisfied.
This is why Lubin's answer drifted toward regulation rather than product. He is not describing a market that lost its appetite for tokens. He is describing a structure in which a consumer fintech cannot fund itself twice.
I ran into a version of this in 2021, screening 50 NFT collections for a fund. One had genuinely beautiful generative work — mathematically elegant, tight, deliberate. The contract beneath it used a proxy pattern that let marketplaces route around creator royalties entirely. The art was real. The architecture underneath it was built to make an obligation disappear from view. I passed, and my reasoning was not aesthetic. When a structure is beautiful and the obligation is missing, the beauty is the mechanism.
A corporate split that renames the shell and moves the protocols is the same shape. Follow the obligations, not the org chart.
Finally, the product story inside the consumer entity deserves its own audit, and it is not the story being told. "Open Money" means card issuance, savings, and trading inside the wallet. Every one of those features is a new signing surface — a new contract to approve, a new delegation to scope, a new message-format assumption to get wrong. Custody language like "users still control their own keys" is true precisely as long as the signing path is not abstracted away from the user, and each feature added to that path is a place where the guarantee thins.
In 2024 I led a review of an AI-agent marketplace that let autonomous agents transact against Ethereum contracts. The finding was a prompt-injection vector that bypassed access controls and exposed roughly $10 million in assets. The lesson generalized far beyond that codebase: the interface is where trust gets transferred, and every convenience feature is a transfer. A wallet that adds a card and a savings product is not the same trust surface as a wallet that only signs transactions, even if the keys never move.
Contrarian
Give the bulls their due, because two of their points are correct and I have not seen them made carefully.
The first is that splitting a conglomerate is the standard, boring, effective remedy for a conglomerate discount. This is not a distress signal. Companies do this when one segment is priced like a liability by association. Lubin's claim that the consumer side compounds faster is plausible, and the decision to give the MetaMask name to the consumer entity and the Consensys name to the infrastructure business is honestly made. Consensys was always the protocol-facing identity. Letting it travel with the protocols while the wallet takes the consumer brand is the naming decision a founder makes when he is being straight about which business carries the original mission.
The second is operational: no migration, no re-issuance, no forced re-verification. Wallet holders keep the same app, keys, and funds. That is the correct way to execute this, and it is worth naming because the last cycle produced several "restructurings" in which user assets were quietly the collateral for a corporate maneuver.
And a genuinely contrarian read on the token: the split may make a MASK issuance more likely, not less. A pure consumer company can ship a loyalty instrument framed as a token without dragging a proving-system business into the securities analysis. The catch is that such an instrument would be non-transferable, non-yield-bearing, and divorced from the cash flow traders have been pricing in for four years. The token arrives. It just will not be the asset anyone modeled.
Takeaway
Watch the related-party contracts, not the org chart. The separation closes by the end of 2026, and by then a MetaMask prospectus will have to disclose what it pays for RPC, what happens to that rate if the counterparty is repriced, and who owns the fee switch. Silence is the only honest consensus mechanism, and right now the silence on the token is the most informative thing in the filing that does not exist yet.
