Ask what you can actually do with crypto and you get two standard answers: “everything” — says the marketing. And “nothing but gambling” — says the pub. Both are lazy, because both avoid the same work: looking it up.

This post is the map of the real. What is documentably being used in 2026 — by whom, at what scale, with what sources? And what was just as documentably discontinued, wound down, or collapsed? Both belong on the same map, or it isn’t one. Two rules up front: every mutable number carries an as-of date, because this post — like everything here — keeps being updated. And where a figure comes from the company itself, it says so. A self-report is not a measurement.

A person paying with a smartphone at an evening street food stall, warm light, the screens unreadable

Illustrative image, AI-generated.

Payments: the stablecoin rail

The by far largest real use case is unspectacular: sending digital dollars from A to B. Visa’s Onchain Analytics dashboard shows an adjusted stablecoin transaction volume of around $1.79 trillion for June 2026 — more than double June 2025; summed over twelve months, around $10.2 trillion. About two thirds of it runs over USDC, just under a third over USDT (as of June 2026).

The word “adjusted” is the actual content here. Visa filters the raw volume twice: per transaction, only the largest single transfer counts (removing technical intermediate hops), and addresses with over 1,000 transactions or over $10 million in volume within 30 days are excluded as bots. How much that removes, Visa’s crypto lead Cuy Sheffield quantified himself in 2024: of roughly $2.65 trillion in raw 30-day volume, about $265 billion remained after adjustment — Visa classified around 90 percent as non-organic. Critics counter that the filters also throw out legitimate market-making; the truth sits somewhere in between. And for completeness: Visa is not a neutral observer here — it runs the dashboard and is itself active in the stablecoin business. The numbers are still the best publicly available — you should just know who publishes them.

Where this has concretely arrived in everyday life: PayPal. Its dollar stablecoin PYUSD has run since August 2023 (issued by the regulated Paxos Trust, with monthly reserve reports), and since July 2025, US merchants can accept payments from external wallets via “Pay with Crypto” — the merchant automatically receives fiat or PYUSD. The advertised 0.99 percent service fee, however, is an introductory price expiring at the end of July 2026; after that it’s 1.5 percent. Still below typical international card fees — but the gap shrinks once the promotional sign comes down.

One sentence has to stand in this section regardless: “stablecoin” is a design claim, not a safety guarantee. Terra/UST erased over $40 billion in a week in 2022 — the mechanics of that collapse are in the bear market post. Today’s large stablecoins are built differently (custodied reserves instead of an algorithm), but the word alone audits no reserve.

Remittances

The use case with the clearest documented benefit is also the least glamorous: remittances — money that migrant workers send home. According to the World Bank’s “Remittance Prices Worldwide”, a classic $200 transfer cost a global average of 6.36 percent in the third quarter of 2025; banks were the most expensive channel at nearly 15 percent. The UN sustainability goal demands a maximum of 3 percent by 2030.

On well-built corridors, the stablecoin rail undercuts that clearly: the Mexican exchange Bitso states it processed around $6.5 billion in crypto remittances on the US–Mexico corridor in 2024 — by its own account roughly ten percent of the total corridor, at costs below one percent. That’s a company figure, not independently audited, but the order of magnitude is plausibly documented.

Two honesties belong next to it. First: the fair comparison is digital versus digital — classic digital services are also markedly cheaper than the 6.36 percent aggregate, which counts bank counters and cash agents. Second: the blockchain leg of a stablecoin transfer is fast and cheap — but the savings can be lost at the fiat ends: whoever wants to cash out pesos needs an off-ramp, and in small corridors without built-out infrastructure, its fees and FX spreads eat the advantage partly or entirely. The rail is cheaper where it is liquid — not everywhere it technically works.

Where this is everyday life — and where it was a state project

So who uses all this? Mostly not the countries that talk loudest about it. In the Chainalysis Global Crypto Adoption Index 2025 (September 2025, data through June 2025), India leads for the third time in a row, ahead of the US, Pakistan, Vietnam and Brazil — the top is dominated by emerging and high-inflation countries, and on-chain activity grew fastest in Asia-Pacific (+69% year over year), Latin America (+63%) and Sub-Saharan Africa (+52%).

What people there do with it is well documented: hedge against their own currency. In Argentina, stablecoins accounted for more than half of all peso crypto purchases between July 2024 and June 2025, per Chainalysis — savings in dollar tokens, because the peso eats them otherwise. In Türkiye, stablecoin purchases reached around 4.3 percent of GDP as early as April 2023 to March 2024 — the highest value worldwide, against a backdrop of inflation that at times ran around 60 percent. That’s not investment behavior; that’s self-protection by other means.

And then there’s El Salvador — the only country that ever made Bitcoin legal tender by law, and therefore the most important real-world test. The honest balance: it failed, documentably. Real usage numbers stayed low throughout — a business survey from late 2022 found that around 98 percent of firms had not made a single sales transaction in Bitcoin, and in polls two thirds of the population considered the project a failure. In January 2025, as part of a $1.4 billion IMF program, congress abolished mandatory acceptance: accepting Bitcoin has been voluntary since, taxes can no longer be paid in BTC, and the state withdrew from its own wallet. The situation since is curious: the government keeps communicating daily Bitcoin purchases — holdings per trackers around 7,700 BTC, worth around $475 million at mid-July 2026 prices — while the IMF states there have been no net new purchases since the program began, the increases being transfers between wallets. Both accounts stand unresolved side by side — that ambiguity is the balance, as of July 2026.

And in Germany?

Here the picture is less spectacular, and that belongs on the map too. According to the ECB’s SPACE payment study 2024, the share of euro-area consumers who own crypto more than doubled between 2022 and 2024 — above ten percent in 13 of 20 countries surveyed. As a means of payment at the checkout, crypto plays practically no role in the same data. Ownership here means: investment, not paying.

Where payment actually happens, it mostly runs over debit cards from licensed providers that auto-convert to euros at the till — Bitpanda, for instance, was among the first large providers with a BaFin license under MiCAR in early 2025. Direct crypto acceptance at German checkouts remains a niche of individual merchants; no reliable official figure exists, so I won’t invent one.

What many don’t know: for German tax purposes, every crypto payment is a private disposal (§ 23 EStG). If less than a year lies between purchase and payment, the price gain is taxable at your personal income tax rate. There’s an exemption threshold of €1,000 per year across all private disposals combined, and it’s treacherous: one euro above, and the entire gain is taxable. The Federal Ministry of Finance circular from March 2025 explicitly treats spending crypto on goods as a disposal and sets documentation duties. The coffee paid by crypto card triggers this event — and so does paying with a dollar stablecoin: on a 1:1 token the gain is usually tiny, but the documentation duty remains. Not tax advice, just the note that in Germany this rail produces paperwork the advertising doesn’t mention.

And Bitcoin’s own payment rail? The Lightning Network crossed roughly $1.1 billion in monthly volume for the first time in November 2025 — real and growing, but next to the stablecoins’ $1.79 trillion there’s a factor of a thousand in between, and the dominant use case is transfers between exchanges, not the bakery. The micropayment experiments that drove Lightning’s growth in 2023 have themselves faded again. If you want to look up the vocabulary behind all this: the glossary keeps growing.

The institutions are already here

Perhaps the biggest shift of the past two years happened not at checkouts but in back offices. Tokenized US treasuries — classic money-market products whose shares run as tokens on public blockchains — stand at around $15.9 billion across 85 products per rwa.xyz (as of Jul 23, 2026). In early 2024 that was around one billion. The names behind it are not startups: BlackRock’s BUIDL manages $2.5 billion — and is, incidentally, only number two now, behind Circle’s USYC; the oft-quoted phrase “largest tokenized treasury fund” has been wrong since mid-2026. Franklin Templeton, whose fund launched in 2021 as the first US-registered tokenized money market fund, sits at around $2.4 billion across its products.

On July 15, 2026 came a step that would have sounded unthinkable three years ago: Ondo launched tokenized stocks in collaboration with the DTCC — the central US securities settlement house. The underlying securities remain in custody of the Depository Trust Company throughout; what trades is a tokenized claim on them. That’s the most serious construction in the field — and at the same time the yardstick to measure the rest against. Robinhood, for instance, has sold tokenized US stocks to EU customers since June 2025, initially including tokens on the unlisted companies OpenAI and SpaceX — whereupon OpenAI publicly clarified: no equity, no partnership, no approved transfer. Buyers hold a contract on Robinhood’s stake in a special-purpose vehicle — two steps removed from the real share. Kraken’s xStocks have run since June 2025; the July 2026 expansion to stocks from Hong Kong, the UK and South Korea is an announcement via a partner, subject to approvals — not yet a trading launch. In general, for tokenized stocks: usually no voting rights, no shareholder status, issuer risk of the intermediate structure. And the scale: around $1.9 billion in total (as of Jul 23, 2026) — against a classical US stock market beyond $50 trillion, that’s a rounding error with a growth curve.

The banks themselves are building too: JPMorgan’s blockchain platform Kinexys has settled over three trillion dollars by its own account and processes over five billion a day on average (as of April 2026). SWIFT declared its blockchain-based shared ledger ready for use in July 2026 — 17 major banks are piloting cross-border payments with it. Important caveat: the ledger orchestrates payment commitments; final settlement still runs over existing infrastructure — a pilot, not regular operation. And the first US federal law for payment stablecoins, the GENIUS Act, has been signed since July 2025 but is not yet in force a year later: the agencies’ implementing rules existed as drafts at the mid-July 2026 deadline, none final; the law takes effect in January 2027 or 120 days after the final rules. “Passed” and “applies” are two different dates.

For clarity, because it often gets thrown into one pot: the digital euro is not a crypto asset but central bank money — centrally issued, no public blockchain. In July 2026 the ECB selected 36 payment service providers for its pilot project; the pilot starts in the second half of 2027, an issuance decision comes only after the EU regulation (expected 2027), earliest first issuance: 2029. So if someone sells you “digital euro” and “Bitcoin” as the same thing in one sentence, you’re listening to marketing.

Machine networks: DePIN

The most idiosyncratic real application is called DePIN — decentralized physical infrastructure: people run hardware at home (wireless hotspots, dashcams, storage servers) and get paid in tokens for it. What’s interesting is that the product here isn’t money — it’s infrastructure.

A small antenna on a rooftop at dusk, the lights of a city stretching to the horizon behind it

Illustrative image, AI-generated.

The biggest example is Helium: a wireless network built from private hotspots that, at the end of 2025, connected over two million people daily by its own account. Since April 2025 there’s been a commercial agreement with AT&T — its customers connect automatically to over 90,000 Helium hotspots in the US and Mexico; something similar runs with Telefónica in Mexico. That is real, paying demand from a classical corporation to a token network. And yet the second sentence belongs here too: monthly revenue peaked at around $2.5 million in March 2026 — while the HNT token fell to an all-time low in 2026. Usage and token price are two different curves, and whoever justifies one with the other is confusing the map with the territory.

Hivemapper has dashcam drivers map streets — 16 million unique road kilometers by September 2024, per company figures, faster than Google Street View in comparable time. Impressive, but the coverage isn’t independently confirmed, and “driven once” is not “kept current”. And Filecoin, the decentralized storage market, is the category’s most honest teaching piece: network utilization rose to a record 36 percent in the third quarter of 2025 (Messari) — which simultaneously means: nearly two thirds of the once token-subsidized capacity lies idle, and the ratio partly rises because capacity is being dismantled. Token incentives build infrastructure faster than real demand grows into it — that’s the DePIN pattern in one sentence. If you want to run infrastructure yourself, by the way, there’s a smaller and more sensible place to start: your own Bitcoin node.

Side benefits: heat and grid

Two applications arise not on blockchains but next to them — from the physics of mining. In Finland, MARA feeds waste heat from Bitcoin mining into two existing district heating networks; the independent reporting on it (Grist, February 2026) puts the residents served at around 80,000 — and delivers the counter-position in the same text: mining heats only one-to-one like an immersion heater, heat pumps deliver a multiple per kilowatt-hour, and additional mining power demand can keep fossil plants running longer. Both are true: the heat is real, and the question of whether it should be bought this way is legitimate.

In Texas, meanwhile, miners get paid for switching off: grid operator ERCOT and the power provider paid Riot Platforms around $31.7 million in credits in August 2023 alone for powering down during the heat wave — more than Riot’s mining earned in the same month. The model continues (for the third quarter of 2025, Riot reported similar magnitudes). Proponents say: flexible large loads stabilize the grid. Critics say: the miners co-create the peak they’re paid to avoid, and the costs land in everyone’s power bill. I’m not arbitrating here — just noting that both positions are documented, and the story “mining saves the grid” is incomplete without its second half.

The counter-ledger

A map that only shows the walkable paths is advertising. Here are the paths that no longer exist.

Play-to-earn. Axie Infinity was the most-used blockchain game in history in 2021 — at the peak, developer Sky Mavis reported around 2.7 million daily active players (a self-report whose exact counting method the sources don’t settle). In the Philippines, “scholars” — players borrowing game characters from investors — earned real money with it for a while. The model tipped before the token crashed: as early as November 2021, the typical daily earnings fell below the local minimum wage of roughly seven dollars outside Manila — the tokenomics inflated rewards faster than new players arrived. Then the reward token fell over 99 percent, in March 2022 North Korea’s Lazarus Group stole around $615 million from the ecosystem’s Ronin bridge, and the scholars left the game en masse — many in debt, having borrowed their starting capital. By mid-2026, trackers estimate daily users in the tens of thousands, roughly 97 to 98 percent below the peak. “Fighting poverty through gaming” became debt and exodus.

NFTs as investment. Trading volume collapsed from around $4 billion in the second quarter of 2024 to $823 million in the second quarter of 2025 (DappRadar); art NFTs collapsed from $2.9 billion in annual volume in 2021 to $23.8 million in the first quarter of 2025 — roughly minus 93 percent. The nuance, in fairness: the market isn’t entirely dead, it has banalized — in the third quarter of 2025, more NFTs changed hands than ever, but at very low prices; as a cheap collectible and ticketing object the technology lives on. What failed is not the file but the 2021 investment narrative. The corporate utility promises went with it: Starbucks ended its NFT loyalty program Odyssey in March 2024 after 15 months of beta, Nike wound down its 2021-acquired NFT brand RTFKT and sold the remains — and at one point the images of over 19,500 CloneX NFTs were temporarily unavailable because a central hosting contract expired. That incident is the whole lesson in one image: token on the chain, image on a web server — the decentralization was often merely claimed.

Corporate blockchains and corporate money. IBM and Maersk shut down their trade platform TradeLens at the end of 2022 — official reasoning: the necessary industry-wide cooperation never materialized. Competing shipping lines didn’t want their data on a platform co-owned by a competitor — the technology worked, the governance didn’t. And Facebook’s world-currency project Libra/Diem was buried in January 2022, its assets sold for around $182 million; the regulatory resistance was insurmountable. History’s irony: precisely that failure paved the way for the stablecoin regulation under which PayPal today legally does what Facebook wasn’t allowed to.

Does this need a blockchain?

Behind almost every entry in the counter-ledger sits the same never-asked question. It has long existed in citable form: US standards agency NIST published a decision flowchart by which a blockchain only makes sense when multiple parties who do not trust each other must write to a shared, immutable dataset without an intermediary — otherwise a database is the better solution. The most-cited academic scheme (Wüst/Gervais, ETH Zürich) reaches the same conclusion. And security researcher Bruce Schneier has argued since 2019 that blockchains don’t eliminate trust, they merely shift it — from institutions to code, exchanges and wallets, which are fallible in turn. These are documented skeptic positions, not editorial opinion — but it’s striking that TradeLens, Odyssey and Diem failed at exactly the spots these three texts had marked in advance. The applications that run — stablecoins, tokenized treasuries, DePIN — plausibly pass the NIST test, by the way: many parties, no shared trust, shared dataset. That’s no coincidence.

How many actually use this?

Finally, the most uncomfortable number. Crypto investor a16z estimates in its State of Crypto Report 2025: around 716 million people worldwide own crypto — but only about 40 to 70 million use it actively each month. Fewer than one in ten. The same report calculates that of raw stablecoin volume, only about a fifth remains after bot adjustment — and that even this adjusted volume clearly exceeds established payment networks. Both halves of the sentence are true: the raw numbers are massively inflated, and the real core is large. (a16z is a party here, as a crypto investor — that this report of all reports states the owner-user gap so honestly rather strengthens it as evidence.)

The soberest research finding fits alongside: a BIS study of exchange apps across 95 countries shows new users arrive above all when the price rises — the pattern of speculation, not payment use. An estimated three quarters of retail users likely lost money on their original investment. And the institutional counter-position to crypto euphoria is just as documented: the Bank for International Settlements judges that stablecoins don’t qualify as a foundation of the monetary system and sees the future in tokenized central bank money; the IMF warns of dollarization of smaller economies and run scenarios. Who ends up right, I don’t know — both sides stand here side by side on purpose, because that’s exactly what the honest state of 2026 looks like: the technology is being used, and the architecture question is open.

The frame all of this happens in

For everyone who wants to use any of this, the frame: in the EU, the crypto regulation MiCAR has been fully applicable since the end of 2024, and the last transitional period for legacy providers ended on July 1, 2026 — since then, only licensed providers may offer crypto services; around 300 are in the ESMA register (as of mid-July 2026). Checking that register before using a provider is basic hygiene. Since the end of 2024, the travel rule also applies: licensed providers must attach sender and recipient information to every transfer, with no de-minimis threshold — real crypto use in the EU in 2026 is traceable like a bank transfer, not the wild west of 2021. And one sentence on custody, because it belongs in every application debate: whoever leaves crypto with a provider holds that provider’s promise — FTX showed in 2022 what that can be worth in the worst case. MiCAR now regulates custody; the trade-off between self-custody and a provider still remains yours.

The map at a glance

That’s the state of things. No revolution, no nothing — but a technology that measurably works in a few unglamorous niches, failed in many glamorous ones, and whose biggest users are now called banks. Dealing with it takes one thing above all: the calm to distinguish documented use from narrated future. That sobriety — using what demonstrably works and letting the rest go — happens to be the same stance the Crypto Collection stands for: no promise, a stance.

Sources and limits

The figures in this post come, wherever possible, from primary sources: Visa Onchain Analytics (stablecoin volumes), the World Bank’s Remittance Prices Worldwide, Chainalysis, the ECB (SPACE, digital euro), rwa.xyz (tokenization), primary releases by Ondo/DTCC, SWIFT, JPMorgan and MARA, IMF country reports on El Salvador, DappRadar (NFTs), Messari (Filecoin, Helium), NIST, BIS and IMF, ESMA, and legal texts (§ 23 EStG, the BMF circular, MiCAR, the Transfer of Funds Regulation). Company self-reports (Bitso, Helium, Hivemapper, Sky Mavis, MARA, Kinexys, a16z) are labeled as such in the text — documented, but not independently audited. Fast-aging values — the Visa dashboard, the rwa.xyz volumes, the value of El Salvador’s Bitcoin holdings, the PayPal fee after the introductory period, the CASP count in the ESMA register — carry as-of dates and will be refreshed as this post is updated. And to take with you one last time: that something is being used doesn’t mean you should buy it. Usage is a fact, price expectation is a wish — this post documents only the former.

Questions, or spotted a mistake? Write to me.