In July 2025, Galaxy Digital sold more than 80,000 bitcoin for an unnamed early investor. The position was worth more than $9 billion, and Galaxy described the transaction as part of the investor’s broader estate-planning strategy.

There is something revealing about that detail.

Bitcoin began as a way to hold and transfer value outside the conventional financial system. One of its earliest great fortunes eventually became something to be distributed across generations, managed through legal structures, and integrated into the ordinary machinery of wealth.

The technology did not fail by becoming part of that machinery. It survived long enough to encounter it.

Most discussions about crypto cycles focus on price. That is understandable because price is the most visible part of the system, and for much of crypto’s history it has been the variable around which everything else reorganised. Developers arrived when capital was available. New users appeared when prices were rising. Narratives hardened during bull markets and disappeared during bear markets. The four-year rhythm became both a market structure and a cultural one.

But price may now be becoming a less useful way to divide crypto’s history.

A deeper change is happening underneath it.

The people who own the assets are changing. The institutions that intermediate them are changing. The most important use cases are changing. Even the arguments used to justify why crypto matters are changing.

That does not mean the old system is disappearing.

It means the centre of gravity is moving.

It is tempting to divide crypto neatly into eras. Bitcoin, then ICOs, then DeFi, then NFTs, then institutions, then AI. The categories make the history easier to remember, but they can also obscure what actually persists.

Technological systems rarely replace one another cleanly.

Mainframes still exist. Email did not eliminate letters. The web did not eliminate television. Cloud computing did not eliminate personal computers. Each new layer changes where the important economic activity occurs while leaving much of the previous layer intact.

Crypto seems to be following the same pattern.

Hyperliquid can thrive at the same time that BlackRock accumulates bitcoin. Stablecoin settlement can expand while memecoin markets remain active. Prediction markets can become large businesses while tokenised Treasury funds are quietly integrated into institutional portfolios.

These things are not contradictory.

They are evidence that epochs overlap.

A useful definition of an epoch is therefore not a period in which one set of behaviours disappears and another begins. It is a period in which the marginal owner, user, institution and source of legitimacy begin operating according to a different logic.

Seen this way, the transition now taking place is easier to understand.

The first broad crypto cohort was largely composed of individuals. Miners, developers, founders, traders, early venture investors and technically curious outsiders accumulated assets directly. The user interacted deliberately with crypto. The wallet mattered. The private key mattered. The decision to participate was often conscious and ideological.

The strongest arguments for the system reflected that origin.

Self-custody mattered because institutions could fail.

Censorship resistance mattered because access could be withdrawn.

Privacy mattered because information asymmetry between individuals and institutions had steadily narrowed.

Permissionlessness mattered because financial systems tend to protect incumbents.

Those concerns have not gone away.

But they are no longer the only reasons capital is entering the system.

A pension fund does not buy bitcoin because it read the cypherpunk mailing list. A fintech does not integrate stablecoins because it wants to weaken the nation state. An asset manager does not tokenise a Treasury fund because it wants to escape regulated finance.

They participate because settlement is faster, collateral can move continuously, distribution becomes global, balance sheets become more programmable, and new forms of automation become possible.

The technology is increasingly being adopted for reasons that are almost independent of the ideology that produced it.

That is usually what deployment looks like.

The idea of a twelve-year crypto epoch can sound artificial if treated as an extension of the halving cycle.

Three halvings do not create a law of history.

The more interesting mechanism is human.

The first broad market cohort began to form around 2013. Bitcoin had existed for several years by then, but it was only beginning to become economically consequential. Someone entering around that period could experience the 2013 boom and collapse, the 2017 cycle, the 2021 cycle and the 2024–25 expansion while remaining inside the same industry.

That is long enough for something important to happen.

A person can enter a market as an outsider and, over a decade, become part of its institutional memory.

The first cycle teaches enthusiasm.

The second teaches risk.

The third teaches proportion.

People who survive long enough stop evaluating the asset in the same way they did when they entered.

A concentrated position that once represented possibility eventually represents responsibility. The question changes from whether the asset can make someone wealthy to whether it still makes sense for most of that wealth to remain exposed to one system, one market and one set of technical assumptions.

This is where market cycles and human cycles begin to interact.

Old holders selling into strength is not new. It happened in 2013, 2017 and 2021. Long-term holders rationally distribute when liquidity appears.

What matters about 2024 and 2025 is not simply that old coins moved.

It is what was waiting on the other side.

More than 470,000 bitcoin held for at least five years moved during 2025, and the dollar value of long-dormant bitcoin moved across 2024 and 2025 was historically large. Some of that activity was certainly custody migration, estate planning or address reorganisation rather than outright selling, so it would be a mistake to treat every movement as a transfer of ownership.

Still, the buyer composition has changed in a way that did not exist in earlier cycles.

US spot bitcoin exchange-traded products now hold a meaningful share of circulating supply. Public companies own large positions. Asset managers, corporate treasuries and institutional allocators have become persistent participants.

The important change is not that these owners are stronger or more committed.

They are governed differently.

An early holder may sell because the position has become too large, because their circumstances have changed or because their conviction has weakened.

An exchange-traded product sells because investors redeem.

A corporate treasury buys or sells according to financing conditions, balance-sheet constraints and capital-market access.

An institution responds to mandates, volatility thresholds, portfolio construction and risk committees.

The asset has not escaped human behaviour.

It has moved into a different institutional architecture.

That matters because architecture determines incentives, and incentives determine behaviour more reliably than narratives do.

The same pattern is visible in people.

There has been a tendency to describe recent departures from crypto firms and foundations as an exodus. That framing is too clean.

Some people have left. Others have reduced their involvement. But many have simply changed roles.

Kyle Samani stepped away from day-to-day management at Multicoin while remaining involved elsewhere in the ecosystem and exploring adjacent technologies. Ethereum Foundation researchers have moved into new organisations rather than disappearing from Ethereum. Crypto-focused venture firms have broadened into AI, robotics and other frontier technologies.

At the same time, other members of the original cohort remain deeply involved.

Michael Saylor continues accumulating bitcoin. Brian Armstrong is pushing into payments and machine commerce. Paolo Ardoino is extending stablecoin infrastructure. Vitalik Buterin remains engaged in Ethereum’s technical and institutional evolution.

The more useful distinction is not between people who stayed and people who left.

It is between those whose role remained static and those whose role recomputed.

This happens in mature industries.

Founders become allocators.

Developers become stewards.

Traders become investors.

Protocol builders move into tokenised finance.

Crypto venture firms become broader technology institutions.

The accumulated knowledge does not disappear. It migrates to places where it has greater leverage.

This may be the real handoff.

Not one generation abandoning the system, but one generation ceasing to occupy the same place inside it.

Crypto is unusual because much of its intellectual development happened in public.

The early texts were political.

Eric Hughes wrote that privacy was necessary for an open society in the electronic age. John Perry Barlow imagined cyberspace as a domain whose legitimacy did not descend naturally from existing governments.

Satoshi’s contribution was architectural.

Instead of arguing that financial institutions should become more trustworthy, Bitcoin attempted to remove some of the situations in which trust was required.

That distinction mattered.

Technology rarely removes a social problem entirely. More often it changes where the problem is located.

Bitcoin moved trust from intermediaries toward cryptographic verification and economic consensus.

Ethereum moved some of it again, from fixed monetary rules toward programmable systems and social governance.

By the middle of the last decade, the literature had moved from philosophy toward economics. Joel Monegro’s Fat Protocols asked where value would accumulate in blockchain networks. Later debates moved toward governance, capture, decentralisation and legitimacy.

Vitalik Buterin’s writing on legitimacy eventually made explicit something that early crypto sometimes preferred to ignore: cryptography can prove that a key signed a message, but it cannot decide whether the person or institution behind that key has the legitimate authority to shape a system.

The underlying problem had not vanished.

It had moved upward.

The history of crypto can therefore be read as a series of relocations of trust.

Banks to protocols.

Protocols to governance.

Governance to legitimacy.

Each technical solution resolves one dependency and exposes another.

That pattern becomes more important as crypto moves from experimentation into infrastructure.

Carlota Perez’s distinction between installation and deployment is useful here, although crypto should probably not be treated as a technological revolution entirely separate from computing and the internet.

It may be more accurate to view crypto as one component of the broader information revolution: a set of tools for ownership, settlement and coordination that becomes more valuable as economic activity itself becomes increasingly digital.

During installation, large amounts of capital are spent discovering what the technology can do.

Most experiments fail.

Some failures are necessary because no one yet knows which parts of the architecture will matter.

Crypto spent more than a decade doing this.

It built exchanges, wallets, bridges, lending systems, stablecoins, smart-contract platforms, token standards, governance systems, privacy tools, prediction markets and every imaginable variation of digital asset.

Much of it was speculative.

Much of it was redundant.

Some of it became infrastructure.

Deployment begins when the technology becomes useful to people who are not particularly interested in the technology itself.

That transition appears to be underway, but it does not imply that speculation disappears.

Hyperliquid is a useful example.

It is deeply crypto-native. It is self-custodial, speculative and culturally much closer to the first epoch than to institutional tokenisation. Yet it also produces meaningful fees and has created a mechanism that connects economic activity to token value.

Prediction markets tell a similar story.

Polymarket and Kalshi are not evidence that retail speculation has gone away. They show that speculation can become a durable product rather than merely a phase of a bull market.

The same applies to DeFi protocols experimenting with token buybacks and explicit fee capture.

Some of the most promising systems may therefore combine the architectural openness of early crypto with the economic discipline normally associated with mature businesses.

That possibility matters because deployment does not require the casino to close.

It only requires the casino to stop being the entire city.

Ethereum may be the most useful case study for what comes next.

The network remains one of the dominant settlement environments for stablecoins and tokenised real-world assets. Major institutional products have been issued there. Large portions of on-chain financial activity depend on Ethereum and its surrounding ecosystem.

At the same time, ETH as an asset has experienced periods of significant underperformance, and the Ethereum Foundation has undergone meaningful restructuring.

Those two facts can coexist.

A network can become more useful while its token becomes less obviously valuable.

That sounds counterintuitive because much of early crypto investing assumed that usage would naturally accrue to the native asset.

Sometimes it does.

But infrastructure economics are rarely that simple.

A railway can transform a region while the railway company earns poor returns.

The internet can create trillions of dollars of value while TCP/IP captures none of it directly.

A payment network can be essential while merchants, banks and software companies capture most of the economic surplus.

Ethereum may settle enormous quantities of valuable activity while issuers, applications, wallets, sequencers, asset managers and exchanges capture much of the associated value.

This does not mean ETH has no value-capture mechanism.

It means value capture has to be demonstrated rather than inferred from importance.

The distinction is likely to become increasingly important in the next phase of crypto.

Crypto can succeed as infrastructure without every crypto asset succeeding financially.

The two ideas were often treated as equivalent during the installation period because speculation rewarded broad exposure.

Deployment forces them apart.

Stablecoins make this distinction clearer.

Their appeal is almost deliberately unoriginal.

USDT does not ask someone to adopt a new unit of account. It gives them an existing unit of account with different settlement properties.

Tokenised Treasury products do not invent a new financial asset. They change how an existing asset can be held, transferred, composed and integrated into software.

This is a very different form of adoption from the token booms of earlier cycles.

The innovation moves from the object to the rail.

A dollar can now move continuously across borders.

A Treasury position can become programmable collateral.

A fund interest can settle on-chain.

A payment can be initiated by software without passing through the same collection of bilateral integrations that traditional finance requires.

The underlying financial claims remain familiar.

What changes is their architecture.

This may ultimately be more important than inventing thousands of new assets.

An economy does not become digital merely because its securities are represented on a blockchain. The deeper change occurs when ownership, settlement and execution become native functions of software.

At that point the system begins to support participants that traditional financial architecture was never designed for.

This is where AI becomes relevant.

There is a tendency to treat AI and crypto as competing narratives because capital and attention move between them.

That may be the wrong level of analysis.

AI creates software capable of taking increasingly autonomous actions.

Crypto creates infrastructure capable of holding and transferring digital assets according to programmable rules.

Those systems solve different parts of the same problem.

A software agent that wants to purchase data, rent compute, pay another agent or compensate an API encounters an awkward financial environment.

Traditional payments assume legal entities, bank accounts, commercial agreements and relatively large transactions.

Software wants something different.

It wants to discover a service, understand its price, obtain permission, pay for it, verify the result and continue.

Protocols such as x402 are early attempts to make that possible.

The current economic activity is still small and should be treated accordingly. Reported volume differs substantially depending on whether speculative and wash-like activity is included.

The interesting part is not current volume.

It is the architectural fit.

The web has always been good at moving information between machines.

It has never been particularly good at moving value between them.

If software agents become economically meaningful actors, that gap becomes more obvious.

Crypto may prove useful not because AI agents believe in decentralisation, but because bearer assets and programmable settlement fit the way machines operate.

That would represent a subtle but important change in adoption.

The first crypto users had to understand crypto.

The next users may not even know they are using it.

Infrastructure usually matures when the abstraction becomes more important than the implementation.

Machine commerce introduces another problem.

A valid signature does not tell us whether an action should have been taken.

An autonomous agent may possess a key and still lack authority to spend $50,000.

It may be allowed to buy compute but not securities.

It may be permitted to access one dataset but prohibited from revealing another.

It may be acting for a company, a household, another agent or itself.

This introduces a new coordination layer around identity, policy, reputation, privacy and accountability.

The technical problem becomes less about whether a transaction can be executed and more about whether the participant has credible permission to execute it.

That is a harder problem.

Blockspace can be increased.

Permission is relational.

It depends on context, institutions and social agreement.

This is another example of trust being relocated rather than removed.

Cryptography can verify identity claims and enforce spending policies. It can make rules transparent and auditable.

But someone still has to decide which claims matter, which authorities are recognised, how mistakes are corrected and who is allowed to revoke access.

The machine economy does not eliminate governance.

It makes governance executable.

That may prove far more consequential than simply putting an AI agent behind a wallet.

Stablecoins reveal a similar tension on the human side.

In countries with unstable currencies or restrictive financial systems, dollar stablecoins can provide genuine economic agency.

Someone can hold a more stable unit of account without needing access to a US bank.

A business can receive international payments without waiting for correspondent banking.

A family can move value across borders outside the constraints of local banking infrastructure.

For those users, the stablecoin rail is not a dilution of crypto’s sovereignty argument.

It is an expression of it.

But the sovereignty is conditional.

The largest dollar stablecoins are issued by identifiable institutions. Those institutions can freeze assets. They cooperate with governments and law enforcement. They can refuse service.

This produces an unusual architecture.

Access becomes more decentralised while control over the asset remains centralised.

A user may escape one institution only to become dependent on another located in a different jurisdiction.

That is not necessarily worse.

It may be significantly better.

But it is different from censorship-resistant money.

The distinction matters because systems are often judged by their interfaces while power accumulates deeper in the stack.

A wallet may feel permissionless while the asset inside it is not.

A blockchain may be neutral while its most important applications are regulated.

A token may be globally transferable while its issuer retains the ability to invalidate that transfer.

Deployment does not resolve these contradictions.

It makes them operational.

The questions that mattered to the early cypherpunks therefore return in a more practical form.

Not whether institutions should exist, but where their authority should end.

Not whether rules should exist, but who can change them.

Not whether enforcement is useful, but whether the person being enforced against has meaningful alternatives.

Sovereignty is rarely absolute.

In practice it is the preservation of optionality between systems.

There is a tendency in technology markets to romanticise survival.

Simply remaining in an industry for a long time does not create insight.

Sometimes it only creates attachment to assumptions that were once useful.

The valuable form of experience is different.

It is the ability to remember why something was designed a particular way while still being willing to change the design when the environment changes.

People who have lived through several crypto cycles have seen how easily activity can be mistaken for adoption.

They have seen protocols subsidise users who disappeared as soon as incentives ended.

They have seen decentralised systems become dependent on a handful of infrastructure providers.

They have seen governance tokens that governed almost nothing, yield that was actually leverage, and communities that existed only while prices were rising.

They have also seen the opposite failure.

Systems can preserve ideological purity so carefully that almost nobody wants to use them.

The difficult work in a deployment phase is not choosing between those extremes.

It is deciding which constraints deserve to survive.

Privacy may be one.

Credible neutrality may be another.

Open access, portability and self-custody may matter even when most users do not exercise them directly, because the existence of an exit changes the behaviour of the institution providing the service.

This is easy to miss.

Optionality has economic value even when it is rarely used.

A person does not need to withdraw assets from a custodian every day for the right to withdraw them to matter.

A developer does not need to fork a protocol for the possibility of forking to constrain governance.

A user does not need maximum privacy in every transaction for private transaction infrastructure to shape the balance of power.

The legacy of the first crypto epoch may therefore survive less through its culture than through architectural choices that preserve alternatives.

That would be a quieter form of success.

It may also be the more durable one.

The Great Handoff is easy to see as a transfer of wealth.

Old coins move.

Funds restructure.

Founders step back.

Institutions accumulate.

But wealth is only the most visible layer.

What is really being transferred is responsibility for the system.

The first cohort had to prove that these networks could exist.

The next cohort will have to decide what they become when they are no longer novel.

That is a different problem.

Early technologies are constrained mostly by engineering.

Mature technologies are constrained by incentives, institutions and the power relationships that form around them.

The internet eventually became less interesting as a network protocol than as an architecture for organising commerce, speech and attention.

Crypto may follow the same path.

The question will not be whether blockchains work.

It will be which forms of ownership they normalise, which intermediaries they remove, which new ones they create, and how much agency remains with the person or machine at the edge of the network.

This is why the handoff matters.

The first epoch was built by people who consciously chose crypto because the existing system did not give them enough optionality.

The next may be built for people and machines who use the same architecture simply because it is the most efficient way to transact.

If that happens, the original ideology will no longer be necessary for adoption.

But the architecture it produced may matter more than ever.

Technologies rarely preserve the intentions of their founders.

They preserve the incentives encoded into the systems that outlive them.

That is ultimately what the next epoch will test.