Eight terabytes.
That is the client-side footprint of the privacy scheme Misha Komarov described to Unchained on September 25 — a figure his team already compressed down from roughly 300 TB. Shielded transactions land at around 700 vbytes, roughly four times a transparent Bitcoin spend. There is no launch date. The second version of the research is penciled in for February 2026.
Three numbers and a date. That is the entire verifiable surface of Bitcoin PIPEs, and for anyone who prices risk instead of narrative, precision about what was not said matters more than what was.
The mechanism is the interesting part, and it is also the part nobody can check.
The pitch: lock a Bitcoin key inside witness encryption, and the only path to unlocking it is producing a witness — a proof — that satisfies some mathematical statement. Chain observers and miners see a normally shaped transaction. No soft fork. No new chain. No company running the thing. Think Zcash's shielded pool, but parasitized onto Bitcoin's liquidity and security budget instead of built from scratch.
Witness encryption, stated plainly, is a ciphertext bound to a proposition rather than to a key. Whoever can prove the proposition decrypts it. The construction is elegant in the abstract. It is also, by the author's own admission, still experimental, and that admission is the single most credible line in the entire report.
I have watched the same pattern play out before, in a different market. In January 2020 I was running a Python arbitrage bot between Uniswap V2 and Kyber — 4,000 fills a month, $12,000 in profit, and a clean-looking backtest. Then a gas spike hit and I gave back $3,500 in one hour because I had not modeled fee volatility as a state variable. The bot didn't fail. I had simply never asked what the thing looked like under load. That is the question Bitcoin PIPEs has not answered yet, and there is no way to answer it from a press interview.
What the report does contain is a set of parameters that, read together, sketch the shape of the machine. Here is where I land on each.
The 4x fee multiple is not the 4x you think. Fees are feerate times vbytes. Quadrupling vbytes does not quadruple cost in a calm mempool, but privacy demand tends to cluster in the same windows as everything else, and a shielded pool competing for block space during a congestion event prices against a fee curve that widens non-linearly. The author calls the premium "not catastrophic." That phrasing is doing a lot of work. If the pool ever carries meaningful volume, the marginal shielded transaction will clear at whichever feerate clears the block, and the discount for being cheap is exactly the privacy that leaks.
The anonymity set problem is structural, not temporary. The report concedes that privacy improves as usage grows. Logically, that means the earliest users have the weakest privacy — and the earliest users are precisely the people with the strongest privacy need and the sharpest threat model. Zcash has spent a decade fighting this. Its shielded share has crawled through single digits and hovered well under a third of activity at best. Monero solved it by making anonymity mandatory and pool-wide, which cost it exchange support and a permanent place on delisting watchlists. Bitcoin PIPEs has neither lever. It is opt-in, which caps the ceiling, and tokenless, which removes the one subsidy that has ever bootstrapped a shielded pool.
An 8 TB client requirement makes this worse in a way that deserves its own paragraph. The population that can run this software is small. The population that can run it and wants to is smaller. The pool starts its life in the exact configuration where the privacy guarantee is weakest, and there is no mechanism in the design that fixes it.
Eight terabytes implies a common reference string, and a common reference string implies toxic waste. I have spent enough time reading audit reports to be allergic to the word "ceremony." A parameter set that large is not a lightweight SNARK; it is a CRS or a large precomputed table, and CRS generation has a specific failure mode. If the setup was multi-party and one participant retained the toxic waste, that participant can forge unlock proofs and spend coins that were supposed to be locked. Flip the risk around and the other failure mode is just as bad: an implementation bug with no pause switch, no operator, and no emergency recovery leaves locked coins locked forever. The same property that makes the protocol uncensorable makes a mistake unrecoverable.
The bot didn't fail; the market changed rules. Here the rule that changed is that an unaudited, unreviewed cryptographic primitive is now guarding actual bitcoin. There is no paper linked, no preprint number, no conference submission, no public repository. "Plans to reuse auditing tools and undergo repeated audits" implies a codebase exists. It does not imply anyone outside the team has read the assumptions underneath it.
"No company operating it" is a cryptographic virtue and an engineering liability. Someone has to ship client builds, distribute 8 TB of parameters, patch a bug at 3 a.m., answer a security disclosure, and maintain documentation. If that someone does not exist, the practical result is a handful of volunteer mirrors and seed nodes — which is functional centralization wearing a decentralization costume. The 8 TB figure is where this gets uncomfortable. Data at that scale does not replicate itself. Whoever holds the mirrors holds the protocol.
The only clean economic signal in the whole document is miner fee flow. A 4x-fee shielded transaction pays miners, not a protocol treasury. There is no token, which means no unlock overhang, no insider vesting cliff, no securities-law question, and no Ponzi flywheel to unwind. In a market where every announcement ships with a distribution schedule attached, that is genuinely rare. It also means nothing subsidizes adoption. Zcash bought its early shielded usage with a founders' reward plus mining incentives. Strip the token away and the cold-start problem gets harder, not easier.
The regulatory frame that matters is not Howey, it is sanctions. No token means no securities exposure — that is the easy half. The hard half is AML and OFAC. The precedents are specific and recent: Tornado Cash sanctioned in 2022 with developer prosecutions following, Samourai Wallet founders charged in 2024. The common feature in both cases was a tool that degraded traceability of crypto assets. The absence of a legal entity reduces the direct target surface, but enforcement does not evaporate when the target does — it migrates downstream to wallets, frontends, and mining pools. The interesting, under-litigated question is whether a pool that knowingly includes shielded transactions is running an unlicensed money transmission business. There is no case law here. There will be.
My contrarian read is that the "Zcash-like" framing is a rhetorical instrument, not a technical comparison. It invites a reader to price a research proposal against a live chain with a decade of production cryptography and a functioning shielded pool. Those two things are not comparable in maturity, trust model, or liquidity. The blind spot is where the money hides: readers will map a speculative primitive onto a market cap and call it analysis.
What the announcement actually does is serve three non-technical functions: recruiting cryptographers, attracting grant funding, and establishing academic priority. Treat it as a talent and capital signal, not a product signal. I trust the log, not the hype — and right now the log is three parameters and a February date.
The real intellectual contribution is a reframing, and it deserves to be said clearly. Bitcoin has no privacy layer because consensus changes require years of political negotiation through a governance process that has repeatedly stalled. This proposal takes the problem out of that room and drops it into a mathematics department. If the direction holds, it matters far more than the specific protocol. If it does not hold, it joins Confidential Transactions and Mimblewimble-on-Bitcoin in the archive of ambitious ideas that never cleared the bar.
One date governs everything: February 2026. Watch six things. Whether the second version of the research ships on time. Whether the witness encryption construction survives peer review at CRYPTO, Eurocrypt, or ePrint — and critically, whether any independent cryptographer reproduces or contests it. Whether the CRS generation method is disclosed with a verifiable ceremony. Whether a repository appears. Whether mining pools take a public position on including shielded transactions. And whether wallet roadmaps show shielded address support, because Silent Payments' adoption curve tells you exactly how fast Bitcoin wallet ecosystems move on hard changes.
The question was never whether Bitcoin can theoretically do privacy. It is whether assumptions written into an 8 TB parameter file survive contact with a hostile party who has every incentive to find the one proof the designers promised could not exist. Alpha decays faster than the code that finds it. So does a cryptographic assumption nobody has audited.