# Qubic: The Long Version (QC, Qubic Church)

## Qubic: The Long Version — Chapter 14. What the Deploy on April 13, 2027 Means

Source: https://qubic.church/books/qubic-the-long-version/14-what-the-deploy-on-april-13-2027-means
Author: QC · Publisher: Qubic Church · 2026

I’m writing this chapter approximately twelve months before the planned date. By the time you read it, the date will either have passed or still be ahead — the exact relation depends on what month the final edition of this book came out in. Which side of the date your copy ends up on changes how everything below reads. But it doesn’t change what I can write right now. And that “right now” sets the register for the whole chapter: a calm analysis of what’s known at the time of writing, plus three explicitly described outcome scenarios.

The date itself appeared in the public field on December 6, 2025, when Come-fromBeyond wrote a short line in one of the chats: “And ETA for Aigarth was set as April 13, 2027.”

A month later, on January 13, 2026, he expanded it more precisely:

“5 years since Qubic launch is just an ETA. And goal of Aigarth is AI, not AGI. Difference between AI and AGI is same as between a car and a BMW.”

This remark should be read in three layers. First: “just an ETA” — Estimated Time of Arrival, an estimate, not a contract. Not “promised,” not “guaranteed,” not “fixed in code.” An estimate based on what the team sees right now. Second layer: the goal is AI, not AGI. An anti-hype framing, drawn directly: Aigarth doesn’t claim to be a universal artificial intelligence capable of solving any human-level task. The goal is a functional AI that knows how to do what it was trained for. Third layer — the car/BMW analogy. Both a car and a BMW solve the same basic task: moving people. But these are different categories of ambition — not different gradations of the same thing. The Aigarth goal belongs to the first category.

### What technically happens on April 13, 2027

The Aigarth deploy is the transition of the framework, published in July 2025, from test mode into production. That very framework: “This framework will be added into the core to host AIs as smart contracts.”

The structural definition of what exactly is being integrated, CfB formulated as far back as November 25, 2023:

“Aigarth would be a standalone project if it didn’t use Qubic’s UPoW and wasn’t intended to run AI as Qubic smart contracts.”

This sentence pinpoints Aigarth’s place in the architecture. Aigarth isn’t an application on top of Qubic, and it isn’t a service that uses Qubic for payment. Aigarth is structurally tied to two specific mechanics of the network: to UPoW, which in chapter eight was unpacked as a way of forcing every computation to work on two fronts at once. What changes at the moment of deploy. Until this date, the Aigarth research-anddevelopment infrastructure runs on the network’s 676 nodes as part of the mining load: nodes generate and train structures. After deploy, a new layer is added to the network — the ability to host AI as smart contracts, accessible through Qubic’s standard SC mechanics. Applications become callable from contracts using Oracle Machines to feed in external data. This isn’t “flip a switch” — it’s a structural event taking, in the vicinity of the date, hours, days, or weeks. April 13, 2027, is the planned date of the event, not the exact time in seconds.

The concrete scenario after deploy looks like this. A user with QU on their account initiates a smart contract call in which the request goes to an Aigarth AI. The contract receives input data through Oracle Machines (if the task needs it — for example, for the price-prediction tasks CfB mentioned in his July 2025 remark as the nearest use-case for Anna). Aigarth inference is performed by the same set of nodes that performs UPoW, as part of their computational work. The result is written back into the contract state. This is AI inference as another type of computation that the network knows how to perform — alongside DOGE mining, message passing, and transaction validation. When I wrote that for Come-from-Beyond crypto would be infrastructure for Aigarth, and not the other way around — that sounded like a paradox. On April 13, 2027, the paradox stops being a paradox. The network becomes what it was built for. Not “crypto that also does AI.” But “AI for which crypto is the load-bearing structure.” This transition is the book’s main structural lock, and it converges on a single date. Arbitrator deactivation — the second half of the same event April 13, 2027, is a two-part date (or even more?). The first half — the Aigarth deploy, unpacked above. The second — the deactivation of the Arbitrator. These aren’t two different events that happen to coincide on the calendar. They’re two parts of one structural transition, declared on a single date back in 2022.

In chapter five I showed three departures of Come-from-Beyond. NXT, founded in 2013 and abandoned in 2014 after the conflict with Sønstebø. IOTA, co-founded in 2015 and left in 2017–2019. The first iteration of Qubic from 2018, abandoned by 2020. In chapter ten I announced a fourth departure — this time with a calendar date announced in advance and with the mechanism built directly into the network’s source code. Now, twelve months before this date, the method is being applied for the fourth time — and the only one of the four whose moment of application is hardwired into the code in advance.

What Arbitrator deactivation means structurally. Chapter ten unpacked this in detail, including through CfB’s central remark from April 19, 2023:

“In best variant Arbitrator won’t be needed at all. Its only drawback - Aigarth will lose access to miners hardware and UPoW will become PoS.”

The concrete mechanism can be described in three points. First: the Arbitrator’s key is destroyed. The compile-time constant in the source code remains, but a signature from a destroyed key cannot be produced. No one, including the team itself, can use the Arbitrator anymore to change the composition of computor nodes or forcibly intervene in the network’s operation. This is the same logic CfB stated back on July 23, 2024: “Arbitrator can do almost nothing, source code guarantees it” — guarantees from code, not from promises. After deactivation, this code simply works without a live key. Second: the network keeps running without the Arbitrator. This isn’t a shutdown, not a functional failure, not an emergency mode. The fallback role is no longer activatable, that’s all. Chapter ten described the Arbitrator as scaffolding — temporary support; on the date that support is removed, the building stands on its own. Third: UPoW structurally becomes PoS-like — by the precise wording of CfB himself. Without the Arbitrator, the network no longer has an external source of new entry points into the computor list. The list is updated only through internal rotation, based on scoring results and the weekly epoch. Good entry points stay, bad ones drop out, new ones come in only through this internal mechanic. This is a known structural consequence, explicitly stated in the same 2023 remark to which the 2027 discussions return.

The uniqueness of the fourth departure is not in the method itself. The method is the same: build the infrastructure, hand it off to run independently, leave. The uniqueness is that the moment of departure isn’t chosen after the fact, as in the three previous cases, but announced in advance and built into the code five years before the date. A pre-fixed date changes the status of the method: from “I’ll step aside when it becomes possible” to “by such-and-such a date this will be feasible, and I commit to that in advance.” A falsifiable public commitment is a different quality of the same action. It’s worth clarifying what “hardwired into the code” means. The Arbitrator’s compiletime constant isn’t a string in a configuration file that you can change by restarting a node. It’s a value compiled into the binary of every version of qubic-core distributed among computor nodes. To replace the key would require recompiling the network and convincing a sufficient number of nodes to switch to the new version — that is, going through the same quorum mechanism described in chapter six as the network’s way of making decisions. Destroying the key after deploy means that even this possibility disappears: there is no longer a private key capable of producing a valid signature from the constant already sitting in the binaries. The network can’t “bring back” the Arbitrator other than by recompiling itself with a new constant and a new key — and that’s no longer the return of the previous role but the creation of a new mechanism from scratch, requiring quorum agreement. Deactivation in this sense isn’t a turn-off, it’s the destruction of the possibility.

April 13, 2022 — April 13, 2027: five years not by accident The starting point of the Aigarth work itself was an even earlier moment. On February 20, 2024, CfB wrote:

“In December 2021 we ran PoC which demonstrated viability of Aigarth approach.” December 2021 — Aigarth’s Proof-of-Concept, an internal test that gave the team confidence the approach works. This is the first of the milestones listed in chapter twelve; the second — December 4, 2024; the third and fourth — July 2025. From the December 2021 PoC to the planned deploy in April 2027 — approximately sixty-four months. From the real start of mining on April 13, 2022 — exactly sixty months, five years.

This double count matters. Sixty-four months from December 2021 is the period of a full research cycle: from PoC to production. Sixty months from April 13, 2022, is the period during which the network’s infrastructure moved in parallel with the research work. If the PoC set “is it possible,” then April 13, 2022, set “exactly how.” UPoW performance, scoring, computor node rotation through weekly epochs, the Arbitrator’s behavior — all of this was calibrated on the same time scale as the Aigarth research. On the final date they’re meant to meet at one point: the research delivers a productionready framework, and the network delivers compute capacity on which the framework starts to run. Five years is the time both sides need.

Five years isn’t a marketing number. It’s a functional window. On July 31, 2024, CfB described the expectation in these words:

“In 2 or 3 years we get Aigarth.”

Said in the middle of 2024, this window is July 2026 / July 2027. April 13, 2027, falls roughly in the middle of it. That is, when the team says “in two to three years,” this date isn’t an aggressive lower bound and isn’t a backstop upper bound; it’s a point inside the announced range, chosen by calendar symmetry with the real start of mining. What’s needed for the five-year period. First — to pass Aigarth’s key milestones, and they have all already been passed: PoC in December 2021, midpoint in December 2024, breakthrough in July 2025, the GitHub publication on July 22, 2025 (all of this is unpacked in chapter twelve, I won’t repeat). Second — to scale up mining capacity to a level at which AI inference through smart contracts becomes economically meaningful for miners. This is the second half of the five-year window, and the next section is about it.

And one more parallel worth saying out loud, because it sits behind the words “five years” and explains a lot. On April 13, 2022, at the moment of mining start, much of the code didn’t yet exist. The same may be true for the final date — the optimizations for compute capacity are being written right now, at the moment this chapter is being written. That doesn’t mean the plan is in danger. It means that in 2022 and in 2027 alike, the work closes precisely on the date — and in the first case that worked out, while in the second it’s so far only planned.

In chapter one I compared the ambition of Qubic with Darby’s ambition in 1709. Coalbrookdale produced the first tons of coke iron then — but the broad adoption of the method in industry took decades. It’s the same here. On the verification date we’ll find out whether a claim of that scale is justified. And the answer doesn’t have to be binary. It can be “it worked, but a year later.” It can be “it half-worked.” It can be “it worked in an unexpected form.” Any of these variants is valid, unlike the falsely binary “success / failure.” A falsifier for an ambition of this scale is always more complicated than “yes / no.”

### What “happened” and “didn’t happen” mean

Scenario A. The deploy succeeds and Aigarth is active. On the date, plus or minus a few days, the framework moves into production. Anna, Maria, and subsequent AIs run as smart contracts, accessible through Qubic’s standard SC mechanics with Oracle Machines for feeding in external data. The Arbitrator’s key is destroyed, the event publicly confirmed. The network keeps running without it — the computor list is updated through UPoW, scoring, and weekly-epoch rotation. What this means for the book’s framing: chapter four is closed empirically — crypto is observably working as infrastructure for AI. What it doesn’t mean: AGI hasn’t been created (and that’s already clarified by the February car/BMW analogy), not all of Aigarth’s technical tasks are solved, the first AI applications may remain quite limited. The book in this scenario hasn’t “won the argument.” It registered a falsifiable framework, and the framework was confirmed — that’s a different quality of result.

Scenario B. The deploy went through technically, but Aigarth isn’t showing what was claimed. The framework works; smart contracts that host AI are technically functional. But AI applications are weak or absent in production: the network’s compute capacity isn’t sufficient for meaningful inference. The Arbitrator’s key is destroyed all the same — that’s a structural event hardwired into the code and happens independently of how active the Aigarth applications are. What this means for the book’s framing: chapter four is closed partially — the infrastructure exists and works, but Aigarth as a research result hasn’t yet matured into a production-ready AI. This is a valid scientific result: the paradigm works at the infrastructure level, and the question shifts from feasibility — that was closed by CfB himself back in February 2026 — to economic scale. And that’s a question the book begins to answer in the next chapter.

Scenario C. The deploy is shifted. On the planned date, the event doesn’t happen as announced. Possible reasons: a technical blocker (blocking bugs were found in the framework, in the core, or in scoring); compute insufficient — the team deliberately postpones because there’s no point in deploying without enough capacity; operational coordination requires another quarter or two of development. The Arbitrator’s deactivation in that case either happens on plan — a structural event in the code, not tied to the success of the Aigarth deploy — or also shifts, which requires a team decision. It’s possible a new ETA is announced — along the lines of the January 2026 remark that five years from Qubic’s launch is, after all, an estimate and not a deadline. What this means for the book’s framing: chapter four remains an open question, carried over to a new date. The falsifier from chapter one partially fired: the date passed, the event didn’t, and that too is a valid observation. A shift of a year or two is standard-scale for infrastructural transitions: the first tons of coke iron in 1709 weren’t produced on the first try either, and broad adoption of the method took decades. A calendar date sets an expectation, but doesn’t determine the answer.

What’s important to say explicitly about these three scenarios. They are not probability distributions. I’m not writing “A is most likely; B less likely; C unlikely.” I’m not predicting the outcome. I’m registering a falsifiable framework: on the verification date, one of these three scenarios will be observed (or a close combination of them), and each gives information.

Scenario A is “Qubic’s victory.” It’s the observation that the infrastructural claim of the book matched infrastructural reality on the verification date. Scenario B is not “Aigarth’s failure.” It’s the observation that the infrastructure works, while the research result hasn’t yet matured to production quality — and that shifts the focus of the next round of discussions from feasibility to economic scale. Scenario C is not “deception of the reader.” It’s the observation that the engineering side requires more time than estimated — and that doesn’t refute CfB’s February 2026 feasibility claim; it remains in force for a new ETA.

I emphasize this because in discussions of crypto projects another framing often operates: either the project “won” or it “failed,” and both verdicts are made emotionally, without structural analysis. That framing is unfit for a scientific posing. Aigarth is a research task, and any research task has several valid outcomes. A falsifiable framework is a way of stating them in advance, rather than reactively picking a frame to fit the result.

In February 2026, CfB stated a conditional commitment: if AI is at all possible within Earth’s current energy budget, Aigarth will achieve it — and the team sees the cards it’s playing as sufficient. That formulation is precise as a falsifier. Either AI is possible within Earth, and then Aigarth is a serious candidate for a path to it. Or it isn’t, and then the question shifts from the paradigm to physics. The falsifier approach doesn’t promise a result — it promises that any outcome will give an answer.

I’m writing this chapter before the date. By the time you read it, the date has either passed or is still ahead — that was already said at the start of the chapter, and I come back to it deliberately. If it has passed — you have the result, and every reader can compare it with the three scenarios above. If it’s still ahead — you have a reason to wait. The book is honest in either case, because it doesn’t claim to know the outcome. A falsifiable framework isn’t a way to win an argument. It’s a way of agreeing in advance that any outcome will yield information.

The main risk — compute capacity between April 2026 and April 2027 The main risk for Aigarth, in CfB’s words: “Aigarth depends on miners” (November 12, 2025) and “Qubic is doomed to succeed. Aigarth? Depends on miners.” (February 7, 2026). It’s not a technical dead end, not a scientific gap, not regulatory pressure. It’s a question of network economics: how much mining capacity will plug into Aigarth tasks by the moment of deploy.

To grasp the scale of what has to happen in the twelve months before April 2027, there’s a concrete number. On January 20, 2026, CfB wrote in a discussion of internal tests:

“We are mining new Aigarth algo tomorrow, 10x bigger than Doge, but not obvious.” Ten times bigger than DOGE mining. DOGE mining is Phase 2 of the network’s current configuration, on which a significant share of mining capacity runs at the time this chapter is being written. The Aigarth task is an order of magnitude bigger. This isn’t a casual comparison; it’s a concrete technical parameter, stated in the discussion of a specific test. And it’s exactly that number that sets the scale of the ramp-up: the network performing X compute on DOGE today must, by April 2027, be able to do roughly ten X on Aigarth.

In chapter eight I unpacked UPoW as a way of forcing one computation to work on two fronts at once. After the Aigarth deploy, the same principle is applied to a new pair of tasks: the compute that today is directed toward DOGE mining as part of Phase 3, after deploy, is redirected to Aigarth inference with the same economic incentives. Structurally, that is, the network is ready — UPoW supports multitasking computation. The question is whether miners will choose such a redirection, and how many of them will plug in by the moment when the choice becomes meaningful.

It’s worth clarifying here what depends on the miners and what doesn’t. What doesn’t depend — the very mechanics of how Aigarth tasks are counted in scoring and how rewards are accrued for them: that’s part of the protocol, published in the github.com/qubic/core repository and not subject to unilateral correction. What does depend — every miner’s economic choice: whether to switch their GPU/CPU power from DOGE payout to Aigarth payout, and in what proportion. That choice depends on the ratio of reward levels between the two tasks, which the network produces every epoch, and on miners’ expectations of how that ratio will move further. The Aigarth team controls neither the first parameter directly (it’s a function of many sides’ participation) nor the second (it depends on individual strategies). What the team does control — the technical readiness of the framework for deploy and the quality of AI applications that the framework will host. Everything else is an external parameter, and that’s what CfB designates with the words “depends on miners.”

The main risk for Aigarth is not feasibility (that, as I said above, CfB closed in February 2026). The main risk is scaling the network to a level at which Aigarth inference becomes economically meaningful for miners. That risk isn’t solved by science or engineering — it’s solved by network economics. And there’s already a partial answer to that question, from August 2025.

### Between plan and proof

I unpacked Axelrod’s three conditions as applied to research work itself — iteration, memory, transparency. The book iterates (its chapters get rewritten). The book holds memory (the hook tracker records every claim still waiting to be closed). The book is transparent (it announces its dates and scenarios in advance, before it can verify the answer). A falsifiable framework is, in essence, the same three conditions applied to its own narrative.

On May 14, 2024, CfB wrote on Twitter:

“I have realized that Artificial Intelligence will not be created, it will emerge. With help of Qubic miners.”

“Will emerge” is the verb of an observer, not a commander. Not “we’ll create AI,” but “AI will emerge — with the help of those who work on the network.” This is the very emergent framing I unpacked: AI as the result of collective mechanics, not as a product designed top-down. And this framing isn’t 2024 marketing; it’s the position the team has worked from since the beginning. Since 2022, when no one outside a narrow circle knew that this emergence would one day surface in the public field; since December 2021, when the first PoC was launched and the team understood that the approach works; since the moment CfB decided that the “build and leave” method could be hardwired into the code in advance, and committed in advance to leaving.

Emergence doesn’t arrive on its own. It arrives if there’s an environment in which it can manifest, and time for that environment to take shape. Earlier chapters described that environment: 676 votes, weekly epochs, UPoW, scoring, smart contracts, governance. Then they described not the environment but the process going on inside it: research, milestones, peer review, artifacts with verifiable properties. Then the moment when the environment and the process must meet at one public point. If CfB is right in his 2024 observation, then on this date the meeting happens — not as an announcement, but as a change in regime in which the network begins to operate. P A r T V The Proof