TechFlowPost published a long-form article on Aug. 24 arguing that crypto in 2026 is being shaped by a simple split: money has come back, but many active participants have not. The piece, written by Fugui, says the market is no longer defined by the same people who powered earlier cycles, even when prices recover.
The article opens with Aug. 19, when Bessent doubled the cap on long-bond buybacks at the Treasury. Yields on 30-year U.S. Treasuries fell from 5.31 to 5.18. Bitcoin then printed three large up days and moved from 64,000 to 78,000, roughly a 20% jump in a matter of days. Calls that the bull market had returned spread quickly.
By the article’s account, capital did return. In the third week of August, U.S. spot Bitcoin and Ether ETFs recorded $2.61 billion in net inflows, the strongest week since last October. Total assets in Bitcoin ETFs recovered to $96 billion, while Ether ETFs stood at $14.3 billion.
People did not fully return with that capital. Over the same period, global monthly active onchain addresses were down 18% year over year, while passive holders were up 16%. The article defines “people” here not as buy-and-hold owners, but as developers, traders and governance participants — the active users who actually use blockchain networks. More people now hold crypto assets, it says, while fewer people use blockchain directly. Monthly active open-source developers across the sector are put at about 28,000, down from a 2022 peak of 45,000.
The article describes this as a collapse in narrative strength despite a rebound in price. Last cycle’s slogan was decentralization; this cycle’s slogan is compliance. Earlier participants talked about private keys and seed phrases. This time, they talk about ETFs. The author argues that if an industry depends too heavily on buy to earn, it risks ending in a Ponzi-like structure because people buy not out of use, but out of the expectation that someone else will pay more later.
First migration: from self-custody to custodial exposure
In the previous bull market, a new user often entered crypto by downloading MetaMask, writing down a 12-word seed phrase and absorbing the ethic that “not your keys, not your coins.”
This cycle looks different. A new participant is more likely to open a brokerage account, type in IBIT and hit buy. Many of them have never generated a private key, do not know what a gas fee is and have never signed a transaction onchain. They own Bitcoin exposure, but may never touch a wallet.
The article puts total U.S. spot crypto ETF assets at $110.3 billion, including $96.07 billion in Bitcoin ETFs. BlackRock’s IBIT is described as the largest single product, accounting for roughly half the category. The ownership mix matters even more: only about 20% of this money comes from institutions that file 13F disclosures, while about 80% comes from retail investors and smaller accounts that do not have to report. In other words, the majority of ETF holders are ordinary investors, not large Wall Street firms.
That crowd is invisible in onchain data. Their purchases do not create new addresses, consume gas or take part in governance. The article says their presence shows up as a rise in holders and a drop in active addresses at the same time. It calls that the core contradiction of crypto in 2026.
It also cites a Bitwise survey of 299 financial advisors. Some 32% said they allocated crypto for clients in 2025, up from 22% in 2024. At the same time, about half of the advisors said 5% or less of their clients actually held crypto. Advisors are learning the product, the article says. Clients are still mostly watching from the sidelines.
The piece argues that this is the first time crypto has had a relatively stable holder base during a bear market. These investors are not watching candlesticks and are not focused on decentralization. They treat crypto as a small position inside a broader portfolio. Their arrival has permanently loosened the link between headcount and onchain activity.
Second migration: from leveraged speculation to limited-risk exposure
The article marks Oct. 10, 2025 as the turning point in the cycle’s user reshuffle. In a single day, more than $19 billion in leveraged crypto positions were liquidated, described as the largest forced liquidation day on record. The trigger came from a macro shock, but the amplification came from market structure inside crypto itself: unified margin systems tied entire portfolios to the weakest asset under stress, while some exchange interfaces froze and left traders unable to exit.
According to the article’s summary of a later review by FTI Consulting, Bitcoin order-book depth on major venues fell by more than 90%, while bid-ask spreads widened from single-digit basis points to double-digit percentages. USDe, a delta-neutral stablecoin, saw one of the most extreme dislocations. On Binance, it briefly traded at $0.60, a 35% discount, while other exchanges still priced it near $1. Because many leveraged products valued collateral using the venue’s own spot price, margin engines marked collateral down and pushed otherwise solvent accounts below maintenance thresholds.
Two months later, open interest was down more than 40% from the October peak. Millions of accounts had been closed, and system-wide leverage had been compressed to roughly 3% of total crypto market capitalization. The article says Bitcoin options open interest exceeded perpetual futures for the first time, pointing to a more defensive position mix and a move away from pure directional bets toward limited-risk exposure.
The author’s reading is blunt: retail traders did not suddenly become more mature. The less mature cohort was removed.
The article then cites a Bank for International Settlements study using data from 95 countries, which found that 73% to 81% of retail investors lost money on their initial investment. It says the causal direction is clear: rising prices draw users in, and as retail buyers chase those moves, the largest holders sell into them and take profits. Roughly 40% of new users fall into the under-35 male demographic, the group identified as most inclined to seek risk.
The boom and fade in memecoins is presented as a clean illustration. Over the past year, more than 13 million memecoins were issued. Solidus Labs analyzed more than 7 million pump.fun tokens and found that 98.6% to 98.7% showed pump-and-dump behavior. Fewer than 2% graduated to Raydium. By September 2025, memecoin issuance had fallen 56% from January.
Speculators still exist, the article says, but their scale is near a multi-year low. It points to an altcoin season index of 39 out of 100 and a fear-and-greed index of 53, categorized as neutral. That combination is described as a classic ebbing-tide signal.
Third migration: from speculative asset to payment tool
Even as prices were cut sharply, one number barely moved: stablecoin market capitalization, which the article puts at about $303 billion. USDT accounts for roughly $183 billion. USDC is put in a range of about $72 billion to $73.7 billion. Together, the two represent around 84% of the stablecoin market. While Bitcoin fell from $126,000 to a level a little above $60,000, stablecoin supply held near the top.
The article calls this the only major user group clearly detached from the price cycle. These users treat stablecoins as a tool rather than an asset. They use them to access dollars, send remittances across borders, defend against local currency depreciation and receive wages. They do not watch price charts, do not take part in governance and may not even know they are using Web3. In practical terms, the article says, they are using a dollar account that happens to run on blockchain rails.
A Castle Island and Brevan Howard survey covering 2,541 users in Brazil, India, Indonesia, Nigeria and Turkey is used to support that case. Some 47% said they used stablecoins for dollar savings, 43% to exchange local currency into dollars and another 43% to get better FX rates. Stablecoins made up more than 10% of assets for 55% of respondents. Nigeria stood out the most: 77% of respondents there held more than 10% of their assets in stablecoins. BVNK’s 2026 data is cited as showing that 59% of crypto-active adults in Nigeria hold USDT, the highest rate globally.
The geography is also telling in the article’s view. Chainalysis data shows onchain value in Asia-Pacific rose 69% year over year to $2.36 trillion, the fastest growth in the world. Latin America rose 63%, and Sub-Saharan Africa rose 52%. Among the top 10 countries in adoption, all but the United States are middle- or lower-income economies. In Brazil, stablecoins account for 90% of onchain crypto activity. The article ties that to World Bank data showing the global average remittance cost at 6.36%, against a United Nations target of 3%.
At the same time, the article warns against reading stablecoin growth as pure utility growth. The same instrument can serve lawful payments and illicit flows. Chainalysis data in the piece shows stablecoins account for 84% of illicit transaction volume, up from 63% in 2024, suggesting that the migration of criminal funds from Bitcoin to stablecoins is complete. The article says those uses need to be assessed separately.
The macro role of stablecoins gets even more attention. As of March 2026, Tether had roughly $141 billion in direct and indirect U.S. Treasury exposure, including $122 billion in directly held short-dated Treasuries. That made Tether the 17th-largest holder of U.S. Treasuries in the world, ahead of Germany, the UAE and South Korea, according to the article. It also cites a BIS working paper arguing that stablecoin reserves concentrated in Treasuries create a direct channel through which global demand for private digital dollar claims becomes demand for U.S. sovereign debt, reinforcing the dollar’s structural role in the international monetary system.
In that framing, stablecoins are no longer just a trading tool inside crypto. They have become a meaningful buyer base for U.S. Treasuries and a vehicle for dollarization in emerging markets.
Fourth migration: from people to machines
The article says stablecoins generate $46 trillion in raw annual transaction volume, but only $9 trillion remains after removing bots and automated addresses. By that measure, about 80% of onchain “activity” is not being done by humans.
It points to the filtering approach used by Visa Onchain Analytics, which removes internal exchange transfers, MEV bots and high-frequency addresses that send more than 1,000 transactions per month or move more than $10 million. Once those addresses are stripped out, only one-fifth of the original volume remains.
Sybil behavior is another part of the picture. In one airdrop, LayerZero removed 803,093 suspected sybil addresses. zkSync had around 6 million unique addresses onchain, but after strict screening, only 695,000 wallets qualified, a pass rate of about 11.6%. The article says it is now normal for one real person to control dozens, hundreds or even thousands of addresses.
AI is accelerating the shift. Chainalysis data cited in the piece shows scam operations tied to AI vendors extracted an average of $3.2 million, 4.5 times more than operations with no AI link. Their median daily revenue rose from $518 to $4,838, while average daily transfer count climbed from 3.89 to 35.1. GitHub Octoverse 2025 is cited as showing AI-generated or AI-assisted code exceeded 40% of code on the platform for the first time.
Payment infrastructure for agents is already being built. Coinbase open-sourced the x402 protocol and formed the x402 Foundation with Cloudflare. Google Cloud and Coinbase also released AP2, or Agent Payments Protocol. Even so, the article says there is still no authoritative public count for AI agent developers or for the volume of onchain transactions initiated by AI agents. It describes this as a new group whose infrastructure exists before its population has been properly measured.
If bots remain the dominant source of onchain activity, the article argues, then user counts inferred from addresses, transactions or TVL stop being demographic measures in any clean sense. It calls crypto the first financial system in which humans are no longer the main acting agents.
Scale and power no longer line up
Looking across all these groups produces a counterintuitive conclusion, the article says. The largest cohort is not the one with the most power. There may be 716 million holders who do not vote and do not build, fewer than 10,000 developers deciding protocol evolution, and a few dozen market makers whose behavior during the October liquidation shock was enough to erase 90% of order-book depth.
On that basis, the article argues that concentration in crypto is not lower than in traditional finance. The concentration just sits somewhere else — not primarily with regulators and banks, but with developers, market makers, large holders and governance delegates, many of whom number only in the thousands.
It backs that claim with a list of concentration metrics: the top 0.01% of entities hold 27% of circulating BTC; DAO governance has a Gini coefficient of 0.998 and a Nakamoto coefficient of 8; the top 10% of NFT traders account for 85% of transactions; $540 million in MEV extraction profits are concentrated in 11,289 addresses; the top three ETF issuers control 89% of the market; and developers with more than two years in the field contribute 70% of code commits. In population terms, the article says, decentralization never really arrived.
Who disappeared, and who stayed
The article points to two visible blanks in the 2026 map of crypto participants.
One is the P2E digital labor force. Axie Infinity reached roughly 2.7 million daily active users at its peak in August 2021, and the Philippines was its biggest player market. Hundreds of thousands of Southeast Asian households once treated play to earn as a primary income source. By the second quarter of 2025, however, wallet counts across the broader Web3 gaming category had fallen 17% quarter over quarter. The article says that group has effectively dissolved.
The other is NFT collectors. Annual NFT trading volume exceeded $23 billion at the 2021 peak. By the second quarter of 2025, a single quarter produced only $867 million in volume. On an annualized basis, that is about an 85% drop from the peak. The article also cites academic work saying the market was concentrated from the beginning, with 10% of traders doing 85% of the activity and 75% of assets selling for less than $15.
Airdrop hunters are not gone, but changing form. The article says the industry has broadly moved toward point systems and pre-token filtering, replacing one-off snapshots with long-term offchain behavior scoring in order to raise the cost of sybil strategies. In effect, what used to be one-time arbitrage has been reshaped into low-paid work that requires sustained effort.
The broader rule, in the author’s view, is that a crowd manufactured by token incentives does not outlast the incentives themselves.
By contrast, groups formed around real demand remain in place even through severe drawdowns. The article singles out stablecoin users because they need dollars, not upside; veteran developers because those with more than two years of experience grew 27% year over year and contributed 70% of code commits; and allocators because banks and long-term holders added during the decline while hedge funds pulled back. The split between trading capital and allocation capital is described as the defining feature of institutional participation in 2026.
Distinguishing between those two kinds of users is, the article says, the most important step when evaluating user numbers for any crypto project.
The human cost
The article argues that crypto cannot be assessed through growth figures alone. It also has to be measured in cost. Chainalysis data cited in the piece shows illicit addresses received at least $154 billion in 2025, up 162% year over year. Of that, $104 billion was tied to sanctions evasion, $17 billion to scams and $3.4 billion to stolen funds. Stablecoins accounted for 84% of illicit transaction volume.
It then contrasts the profile of victims with the profile of active participants. Active participants are described as young men prone to overconfidence. Victims are older people who lose retirement savings to fake investment platforms or to AI-generated videos impersonating their children. FBI data shows fraud losses among Americans aged 60 and over reached $7.7 billion in 2025, up 37% from a year earlier.
The article draws a hard line between investment losses and victimization. When BIS says 73% to 81% of retail investors lose money on their initial investment, that refers to active participants making bad trades. Scam victims lose 100%, the article says, because they were never investing in the first place.
It also discusses involuntary participants. United Nations agencies estimate that at least 120,000 people in Myanmar and around 100,000 in Cambodia are held in online scam compounds, with victims sourced from more than 50 countries. The article says these people are not crypto users in any ordinary sense, but they are part of crypto’s humanitarian cost because crypto serves as the settlement layer for the fraudulent output they are forced to produce.
From a community of people to infrastructure for capital
The article closes by tying all four migrations together. Crypto, it argues, is moving from a community of people to infrastructure for capital. Human participation is shrinking. Machines are gaining ground. Retail is leaving. Institutions are coming in. Speculation is fading in some segments, while practical use is growing in others.
The author does not frame that as purely negative. An industry moving from adolescence into adulthood is bound to lose some illusions. But it also loses some of what once made it interesting. Earlier crypto culture was filled with hackers, punks, coders who believed code was law and forum arguments over technical ideals. Today, the article says, the field is increasingly populated by compliance officers, financial advisors, ETF product managers and stablecoin issuers.
The old question was whether crypto could change the world. The new question is whether it can fit inside an investment portfolio. What came back was capital, not belief.
Even so, the article ends on a qualified note. Stablecoin users in Nigeria receiving wages and developers working on ZK scaling are held up as examples of real problems still worth solving. Those efforts may not be fashionable, and they may not make for easy hype, but they remain a reason for the industry to exist.
Buy to earn may be losing its path. Build to earn, the article says, still has a long road ahead. And AI agents — a “tenth type” of participant that has not yet been properly measured — may already be on that road.
