Open a court opinion from fifteen years ago — a real one, a Supreme Court decision — and start clicking the links in the footnotes. A striking number of them lead nowhere. The page moved. The site shut down. The company that hosted it was acquired, and the new owners had no reason to keep a decade-old file alive.

This has a name. Researchers call it link rot, and it is everywhere. When Harvard researchers checked every web link cited in U.S. Supreme Court opinions, they found that roughly half — about 50 percent — no longer led to the material the justices had cited. In the most important legal documents, the country produces. In some Harvard law journals, more than 70 percent of the links had rotted the same way. Not because the words were wrong. Because the place that held them stopped paying the bill.

We tell ourselves the internet is forever. It is the opposite. The internet is a city built entirely of rented rooms. Everything you have ever posted, uploaded, or cited lives on a hard drive that someone else owns, in a building someone else pays for, under terms someone else can change. The moment the rent stops — a bankruptcy, a policy shift, a quiet decision in a meeting you will never hear about — the room is cleared and your thing is gone.

For a vacation photo, fine. For a legal record, a scientific dataset, a piece of journalism that someone would prefer to disappear, a financial transaction history — "as long as someone keeps paying" is a terrifying foundation.

There is one crypto project whose entire purpose is to fix this. Its promise is a single, audacious word.

Forever.

And the whole story — the technology, the token, the believers, the critics — comes down to one question: can anyone actually keep that promise?

Arweave is a decentralized storage network with one job: store a file permanently, for a single payment, made once.

Not "store it while you keep paying." Not "store it for a year, then renew." Pay one time, today, and your file becomes part of a permanent, public, tamper-proof record that lives across thousands of independent computers around the world — with no company in the middle and no off-switch.

The community calls the result the permaweb: a version of the web where the things you put up simply stay up. You can store a document, a photo, an entire website, or even a working application. Once it is in, it is in. No single government, company, or person can quietly delete it or rewrite it.

Its founder, a computer scientist named Sam Williams, describes what he is building in unusually grand terms: a "decentralized collective memory for humanity." He launched the project in July 2017. The native token that powers it, AR, has a hard cap of 66 million coins — and nearly all of them, roughly 65.65 million, are already in circulation. There is almost no new supply left to create.

That is what. The interesting part is how "pay once, store forever" should be impossible.

Here is the problem with forever. Storage is not free. Every file sitting on a hard drive somewhere costs real money — electricity, hardware, maintenance — every single year, indefinitely. So how can you charge someone once and cover a cost that never ends?

Arweave's answer is borrowed from how great universities fund themselves.

When a donor gives Harvard $10 million, Harvard does not spend $10 million and then go looking for the next donor. It invests the gift and lives off the returns. The principal stays. The scholarship it funds can run, in theory, forever — paid for by a single act of generosity that happened once.

Arweave does the same thing with your storage payment. When you upload a file, only a small portion of your fee is paid immediately to the people who store it. The rest goes into a shared pool — a storage endowment. That pool is designed to keep paying the people who store everyone's data, century after century, drawing down only when it needs to.

And the math has a built-in tailwind: storage gets cheaper over time. It always has — dramatically. The cost of storing a gigabyte has fallen by about 30 percent per year for decades. Arweave doesn't even bet on that. It conservatively assumes storage costs will fall by just half a percent a year — and on that cautious assumption, a single upfront payment is calculated to cover roughly 200 years of storage.

In real numbers, storing a gigabyte permanently costs around $7 — about two-thirds of a cent per megabyte — over two centuries.

That is the elegant idea at the center of Arweave. You are not renting space. You are endowing your data.

To see why people get excited, line Arweave up against the alternatives.

Versus regular cloud — Amazon, Google, Dropbox, iCloud. Those are rentals, full stop. You pay every month, and the day the payments stop, your files are eventually deleted. A company owns the building and holds the only key. Arweave flips the relationship: you pay once, you own your spot, and there is no landlord who can raise the rent or change the locks.

Versus other decentralized storage — Filecoin, Storj, Sia. This is the comparison that actually matters, because these are crypto-native too, and most people assume they all do the same thing. They don't. Filecoin — the biggest of them — is essentially a decentralized rental market. It spreads your data across many independent providers rather than one company, which is more resilient, but you still buy storage in time-limited deals that must be renewed. Stop renewing, and the data can still vanish. It optimizes for cheap, large-scale archives — around nineteen cents per terabyte per month — not permanence.

Arweave made the opposite bet. It is more expensive up front, slower, and not built for huge volumes of throwaway data. What it offers that none of the rental models do is permanence as a default — baked into the protocol, not dependent on anyone choosing to renew.

Different philosophies, not different flavors of the same thing: renew-forever versus pay-once-forever. Arweave is the only one promising you never have to think about it again.

I run every project through the same first filter: The Gate — eight questions, each scored green (strong), yellow (mixed or unproven), or red (weak). A project passes to the deeper dive only with five or more greens and no more than one red. Here's Arweave, honestly:

The tally: 5 green, 3 yellow, 0 red. Arweave passes the Gate — cleanly, on the actual rule.

But notice where the yellows land. All three — token economics, governance of pricing, regulatory permanence — cluster around the same nerve: can the money behind "forever" actually last? The Gate says Arweave is real and worth your attention. The yellows say the entire case lives or dies on one unprovable bet. So that's where the rest of this goes.

Strip Arweave down, and you find one load-bearing assumption holding up the whole structure:

Storage will keep getting cheaper, basically forever — and AR will hold enough value to pay the bills along the way.

If that holds, the endowment doesn't just survive — it gets richer over time, and "forever" is real. If it breaks, the most beautiful promise in crypto becomes something darker: an IOU the network may not be able to honor.

This is exactly where serious people split.

The believers point out how much cushion is built in. Arweave assumes storage costs fall just half a percent a year. The actual historical rate is closer to thirty. As long as reality lands anywhere near the trend of the last forty years, the endowment is not just solvent — it's wildly overfunded, capable of paying for your data not for centuries but for far longer. In their view, the conservatism is the whole point: they intentionally bet on the worst case and still come out fine.

The critics — including Casey Rodarmor, a respected and famously skeptical voice in Bitcoin — make a sharper argument. They call it an unfunded liability. The protocol has promised perpetual storage in exchange for a one-time payment, but it cannot guarantee that future money will cover that promise. Two things have to keep going right, indefinitely: storage costs have to keep falling, and the AR token has to hold enough value to actually pay the people doing the storing. If storage costs ever flatten out, or if AR's price stays depressed for a long stretch, the endowment could run short — and a promise of "forever" that runs out of money was never really forever. It was marketing.

Both sides are arguing about something genuinely unknowable: what the world looks like in 100 years. That is the honest center of this story. Not "is Arweave a scam" — it plainly is not. The real question is whether forever is something a system can promise at all, or whether it is a word that always, eventually, comes with an expiration date nobody wants to print on the box.

Arweave is not a revenue-and-market-share story like a typical startup. There is no subscription line to model. The "math" here is the endowment math — and the token that sits underneath it.

The token. AR is capped at 66 million coins, and about 65.65 million already exist. That near-total supply is a double-edged thing. On one hand, there is almost no future inflation to dilute holders — rare in crypto. On the other, it means the network can no longer lean on freshly minted coins to pay the people storing data. Increasingly, those rewards have to come from real usage and from the endowment itself. The economics have to work, not be subsidized by printing.

The volatility. AR is a crypto asset and it trades like one. It touched roughly $91 in November 2021, then fell to low single digits. As of mid-June 2026 it sits around $2.10, a market cap near $138 million — more than a 40-fold drop from its peak. That swing is not a footnote — it is the risk. The endowment's ability to pay tomorrow's storage bills is denominated, in part, in a token that can lose 90 percent of its value and has.

The tension, in one line. The thing that makes Arweave beautiful — pay once, walk away, trust the endowment — is the same thing that makes it fragile: you are trusting a 200-year financial model denominated in one of the most volatile assets on earth.

TLD

Arweave charges you once — about $7 a gigabyte — and promises to store your file for ~200 years, funded like a university endowment that lives off falling storage costs.

The believers say the safety margin is enormous: it assumes costs fall by 0.5% a year, when they've historically fallen by 30%.

The critics say it's an unfunded promise: if costs ever stop falling, or the token stays cheap, "forever" runs out of money.

Nobody can prove either side, because the argument is about the next hundred years. That uncertainty is the asset, and the risk, at the same time.

The whole thing rests on one assumption. Storage costs must keep declining and AR must hold enough value to fund storage over time. The historical trend strongly favors this. But "strongly favors" is not "guarantees," and the time horizon is centuries.

The token is brutally volatile. AR has gone from ~$91 at its 2021 peak to ~$2 today. Because the endowment's purchasing power is tied to the token, a prolonged, deep price collapse is the scenario that most directly threatens the promise of permanence.

It is expensive and slow by design. Around $7 per gigabyte up front, versus pennies-per-month rentals, and not built for high-speed or high-volume data. Arweave only makes sense when permanence is the thing you actually need. For most ordinary storage, it is the wrong tool.

"Forever" has fine print. Arweave's own ecosystem is careful to say it guarantees data permanence, not network permanence — the data is meant to survive even if the protocol itself someday evolves or migrates. That's an honest distinction, but it means "forever" is a commitment to an outcome, dependent on incentives continuing to work, not a law of physics.

This one is a little different from a typical investment, because Arweave is two things at once: a service you can use, and a token you can hold. Treat them separately.

If you want to use it — to permanently store something that genuinely matters to you — the question is not the token price. It is: do I believe the endowment math holds well enough that this file will outlive me? For high-value, can't-lose-it data, even skeptics concede the odds are reasonable. Just go in understanding that "forever" is a well-reasoned bet, not a certainty.

If you're looking at the token, the usual discipline applies, and harder, because of the volatility:

Size it for total loss. This is a speculative bet that a specific vision of the future will win out. Never spend money you can't afford to lose entirely.

Watch usage, not price. The most honest signal Arweave gives off is the data it stores — it kept climbing even when the token cratered. If real-world usage keeps growing through bad markets, the thesis is alive. If storage growth stalls, the story is breaking, whatever the chart says.

Watch the endowment debate. The unfunded-liability critique is the load-bearing argument against Arweave. If the project ever responds to it — dynamic pricing, stablecoin-denominated rewards, more transparency about the math — that's a meaningful signal that the team takes the central risk as seriously as the critics do.

The honest verdict:

Arweave is not a scam, and it is not vaporware. It does something real that nothing else does, and people are using it for exactly that. The technology works. The usage is growing. The vision is coherent and, frankly, beautiful.

But it has made the boldest promise in this entire space — forever — and that promise rests on an assumption about the next hundred years that no one alive can verify. The believers have history and conservative math on their side. The critics have basic financial honesty on theirs: you cannot fully fund an infinite obligation with a finite, volatile payment and call the matter closed.

Both can be right for a very long time before we ever find out who was right in the end.

That is the whole story. In a world where everything you've ever put online is rented, deletable, and controlled by someone else, Arweave is the one project handing you a key to something permanent — and asking you to trust that the lock will still be there in two hundred years.

Whether you take that key is, fittingly, entirely up to you.

This is a structural analysis. Nothing here is financial advice.