Miro raised at $17.5 billion in January 2022. Bending Spoons just bought it for $1.36 billion in cash. Under four years. A 92% markdown on the undisputed category leader in visual collaboration. Floor price broken. Truth verified.
I pulled that number myself, and that matters, because the trade brief that surfaced this week listed five bullet points, no author, no publish date, and no citation for the original valuation. I spent a working afternoon cross-referencing the C-round figure against financing archives and secondary coverage. The 92% haircut is the only figure that reframes the entire deal, and the source never printed it. Data checked. Community warned. When the wire services won't compute the loss, nobody downstream can price the risk.
Strip the corporate polish and this is distressed arithmetic. Miro's backers — funds that wrote nine-figure checks at the pandemic peak — chose guaranteed cash over a longer, riskier bet on an independent listing. Bending Spoons, the Italian acquirer, does not build products from scratch. It buys mature subscription businesses, compresses costs, raises prices, and harvests free cash flow. WeTransfer. Evernote. Meetup. StreamYard. The pattern is industrial, repeatable, and profitable.
Why now? Because the collaboration boom is over and the bundle war has begun. Figma folds whiteboarding into its design suite at no marginal cost. Microsoft staples Whiteboard into its productivity bundle. Atlassian embeds boards inside Confluence. None of them need whiteboarding as a profit center. Miro needed it as the whole business. Add the AI lag — Miro's generative features shipped behind Figma's and Notion's — and the growth story lost both legs. That asymmetry, not mismanagement, converted a $17.5 billion narrative into a $1.36 billion receipt.
I have seen this exact movie from inside the theater. In April 2021, during the Meebits surge, I embedded with the early collector Discord to verify floor prices against wash-trading bots. Three developers, forty-eight hours, twelve thousand transactions. We built a Python script to flag suspicious wallet clusters and published the results as a public dashboard so new buyers could check history before they bought. What we found was rarely fraud. It was phantom demand — a price with no cash behind it. Miro's valuation carried the same disease, just at institutional scale and with a longer incubation period.
Now connect it. Miro is not a crypto story. It is the template for one.
The mechanism that crushed Miro's multiple is the mechanism now threatening a specific, overcapitalized slice of crypto: standalone infrastructure that sells a feature as a product. The cleanest case is the data availability layer.
Celestia, EigenDA, and their competitors carry valuations that assume every rollup will need dedicated, metered, externally purchased block space. I have audited rollup deployments for years, and the honest number is unglamorous. The overwhelming majority of rollups post so little data that the DA fees they generate would not cover the cost of a coffee. A standalone DA market is a product hunting for a demand curve that mostly does not exist yet. Most rollups are served perfectly well by blobs on Ethereum mainnet or by posting calldata — a feature of the base layer, not a company you build an empire around.
This is Miro's disease in different clothing. Ethereum, like Microsoft, can offer the feature essentially for free because it monetizes something else. A point solution competing against a bundle always loses the margin war, and usually loses the customer too. When the base layer owns the bundle, the point solution isn't acquired at a discount. It's abandoned at par.
There is a second, sharper parallel. Miro's revenue engine was seat-based subscription — you pay per user, per month. That model has a crypto twin: the emission schedule. Both assume continuous expansion. Both look healthy while users are being added and both collapse when net additions stall, because the cost of servicing the base keeps climbing while the new revenue stops. Seat-based SaaS and token emissions share the same failure mode: they reward onboarding and punish retention. Miro hit that wall in 2023. Several modular tokens are hitting it now, disguised as "incentive programs."
Here is the third flag. Bending Spoons's model is cash-flow harvesting, and crypto has its own version: the treasury-management playbook. Token projects mark their balance sheets at the last private round or the last trade. When the narrative that justified that price dies, the treasury is repriced in slow motion — first the secondary market, then the vesting schedule, then the foundation's runway. Miro's investors took four years to accept a 92% haircut. Crypto projects often take four weeks. The repricing is faster here because the mark-to-market is public, continuous, and merciless.
That speed cuts both ways. It exposes overvaluation sooner, but it also lets insiders exit before retail realizes the number was never real. This is where the oracle comparison becomes uncomfortable. DeFi's deepest structural flaw is the latency between what a price feed reports and what the market actually clears at. Chainlink centralized that feed into a permissioned node set and the industry called it decentralization. Miro's sale is the same latency problem in analog form: the printed valuation lagged the real clearing price by years, and everyone who trusted it — option holders, secondary buyers, late-round investors — paid for the gap.
So watch the transmission path into crypto. Institutional allocators who just watched a SaaS leader lose 92% will re-underwrite every "category leader" thesis on their book, digital assets included. The same model, the same meeting, the same spreadsheet. The question they will ask is the one Miro could not answer: if the giant on the other side of the market can bundle your feature for free, what exactly are you selling? Watch how fast the funding follows the bundled outcome.
The answer for a handful of infra teams is real depth — proprietary data, genuinely scarce blockspace, or a network effect that crosses organizational boundaries. The answer for most is theater. Project KYC is the clearest example: buying a few wallet holdings defeats the screening entirely, while honest users absorb the cost and the latency. It is a feature sold to look compliant, not a product that produces compliance. Same category error, different invoice.
The unreported angle is not that Miro was overvalued. Everyone can see that now. The unreported angle is that the bundle war has a crypto front, and almost nobody is pricing it.
Every Web3 infrastructure token whose pitch is "we are the X layer for the modular stack" is a point solution. The modular thesis assumes each function — execution, settlement, consensus, data availability — gets unbundled into a specialized market with its own token. That is a beautiful architecture diagram. It is also the exact structure that loses to vertical integration. Ethereum's roadmap is quietly rebundling everything: blobs for DA, based rollups for sequencing, shared settlement on L1. Each roadmap item deletes a standalone product's business model, and it does so for free.
Analysts still model these tokens as if demand is guaranteed by architecture. Architecture is not demand. Customers buy outcomes, and when the base layer hands out the outcome as a default, the specialized market evaporates. Miro's board learned that in dollars. Crypto's will learn it in token price.
And the media layer that should flag this is failing the same way the Miro brief failed: five bullets, no date, no author, no math. If your source doesn't show the 92%, your source isn't doing analysis. It is doing stenography.
The question to carry into next quarter is not whether Bending Spoons can squeeze cash from Miro. It probably can, for a while. The question is which crypto "category leader" is quietly carrying a printed valuation that the market cleared out from under it months ago. Liquidity gone. Run — or check the order book yourself before someone checks it for you. I will be watching DA fee revenue and rollup data throughput as the first clean tell. When the bundle wins, the feature loses. The only question is who admits it first.