Crypto, Bubbles, and The Coming Agent Economy

Why crypto’s boom-and-bust cycles look less like a terminal bubble and more like the messy monetization phase of a new financial rail.

Crypto has spent most of its life underneath one convenient dismissal: “it’s just a bubble.”

It’s a comforting story if you’re sitting on the sidelines. If it’s a bubble, you get to be right eventually. If it isn’t, you have to explain why a parallel financial system bootstrapped itself in front of you and you stayed in T‑bills.

What’s more interesting than “bubble or not” is a different set of questions: has crypto actually found product–market fit, how does it compare to AI as a technology and capital cycle, and what would it take for crypto to grow non‑linearly in a world filled with AI agents?

This essay is my attempt to answer those, or at least give us a cleaner way to think about them.

What We Actually Mean By A Bubble

Every important technology looks like a bubble from the outside. The pattern is almost boring: a new rail appears, money pours in, stories outpace cash flows, and then gravity reasserts itself.

Minsky and Kindleberger broke this down long before tokens existed.¹ ² First you get displacement — railways, the internet, generative AI, Bitcoin — something that makes the old way of doing things obviously worse. Then a boom as capital and attention chase the upside: higher prices are explained away as “the market finally gets it.” At some point you slide into euphoria: people stop asking what the asset does and start asking only whether someone else will pay more. That’s when the thing itself turns into a vehicle for leverage, storytelling, and career risk. Eventually the people closest to the plumbing step back, you hit profit‑taking, and the rest of the market discovers the narrowness of the exit.

Richard Bernstein’s checklist is the modern risk manager’s shortcut: if you see too much liquidity, too much leverage, universal access, a flood of new issuance and crazy turnover, you’re probably in a bubble, not just a repricing.³

Crypto has hit those conditions before. 2017 and 2021 weren’t “misunderstood value plays.” They were classic displacement‑to‑euphoria arcs sitting on genuinely new rails. What matters now is whether, in 2026, we’re still in that zone — or whether crypto has quietly shifted categories from “pure bubble” to “new infrastructure with periodic blow‑offs.”

A Simple Mental Model For The Crypto Market

To get past vibes, four questions are enough to “health check” the market each cycle:

• Strain — How much capital is going in versus how much revenue is coming out?
• PMF — Are there real users with real jobs-to-be-done that persist through cycles?
• Structure — Who is holding the risk: short‑term tourists, long‑term institutions, or big balance sheets that don’t panic instantly?
• Cycle — Where are we on the boom → euphoria → profit‑taking → panic spectrum right now?

This is loosely inspired by Azeem Azhar’s AI “boom vs bubble” dashboard, where he tracks economic strain, industry strain (revenues vs capex), revenue momentum, valuation heat and funding quality for AI. He lays this out in Exponential View’s “Is AI a bubble?” and in follow‑up discussions of his five‑gauge AI dashboard.⁴

For crypto, we can steal the structure without copying the details.

Has Crypto Actually Found Product–Market Fit?

If you mute the price charts and just watch behavior, one answer is obvious: the world has found product–market fit for dollar rails that do not depend on your local bank.

Stablecoins are the clearest example. a16z’s State of Crypto 2025 report and follow‑up coverage point to tens of trillions of dollars in gross stablecoin transaction volume over the last year, with roughly 9 trillion dollars of “real” adjusted volume in 2025 alone — more than PayPal and in the same conversation as Visa.⁵ ⁶ TRM Labs’ 2025 Crypto Adoption and Stablecoin Usage Report and Chainalysis’ 2025 Global Crypto Adoption Index both show that a lot of this isn’t just wash trading between exchanges; it’s remittances, savings, and payments in places like India, Nigeria, Brazil and Turkey, where the alternative is capital controls and double‑digit inflation.⁷ ⁸

People aren’t routing that kind of volume through stablecoins because it’s funny. They’re doing it because it works better than their banks.

On top of that, DeFi has quietly rebuilt itself. Total value locked has climbed back above the 200 billion‑dollar mark, with late‑2025 industry reports putting DeFi TVL around 237 billion dollars and documenting a strong recovery in lending and derivatives.⁹ Exchange and analytics reports (for example, Phemex and AInvest) now show double‑digit billions in annual protocol revenue across lending markets, perps and DEXs.¹⁰ ¹¹ Stablecoin issuers like Tether, in their own financial statements, report multi‑billion‑dollar profits on reserves.¹²

That is what early, messy product–market fit looks like for financial infrastructure: the casino layer can be way up or way down, but the pipes stay in use and the toll booths keep charging.

If you had to grade it: stablecoins and exchanges are clearly across the PMF line; DeFi “blue chips” are in strong emerging PMF territory; NFTs, gaming, and most new L1s are still in the prove‑you’re‑not‑just‑beta phase.

Is Crypto A Bubble Right Now?

Run the four‑question test.

On strain, the picture is more sober than in previous cycles. Galaxy Digital’s Q4 2025 crypto VC report, for example, estimates on the order of 20 billion dollars of venture capital going into crypto and blockchain across roughly 1,600 deals for the year.¹³ On the other side of the ledger, DeFi revenue reports and stablecoin profit disclosures put 2025 protocol and platform revenue in the mid‑teens of billions, dominated by stablecoins, centralized exchanges, perps and lending markets.¹⁰ ¹¹ ¹² That’s not a perfect match, but it’s at least the same ballpark. This is very different to 2017, when investment and token issuance dwarfed any notion of real cash flows.

On PMF, the key is that usage doesn’t go to zero when price does. Stablecoin supply and volumes compressed in the last drawdown but didn’t collapse. DeFi kept clearing trades. Global ownership counts continued to grow, especially in emerging markets. TRM Labs and Chainalysis both highlight that adoption is now broad‑based across income levels and regions, not just a US/EU retail phenomenon.⁷ ⁸

On structure, we’ve shifted from “your barber is levered long on altcoins” to a more layered holder base. There are spot ETFs and ETPs absorbing institutional and retirement money; corporates and funds experimenting with stablecoin treasury and tokenized treasuries (see Amundi’s Cryptocurrencies Break Into The Mainstream 2025 and Henley & Partners’ Crypto Wealth Report 2025); and banks and fintechs quietly embedding on‑ and off‑ramps.¹⁴ ¹⁵ Retail is still here, but it’s not the only pillar under the market. That matters for how panics play out.

On cycle, 2025 clearly had euphoric pockets: reflexive low‑float tokens, meme seasons, the usual suspects — plenty of exchange and media pieces framed it as “Crypto Bubble 2025?”¹⁶ The subsequent washout has already taken a lot of heat out. Derivatives recaps like the CoinGlass/Pi42 2025 trends report and KuCoin’s perps market analyses show that open interest and leverage were materially flushed into early 2026.¹⁷ ¹⁸ On‑chain analysts tracking valuation metrics like NVT, MVRV and NUPL through dashboards such as Woobull, Bitbo, Glassnode and Bitcoin Magazine Pro have likewise pointed out that those metrics, while elevated, are not at the extremes seen at prior cycle tops.¹⁹ ²⁰

Put together: parts of crypto are perpetually bubbly, but the ecosystem as a whole increasingly resembles a high‑beta growth sector with real revenue and recurring use, not a singular, terminal bubble waiting to pop.

Crypto vs AI: two different risk profiles

Crypto keeps getting yoked to AI in narratives because both attract capital and attention. Under the hood, they’re exposed to different failure modes.

AI is fundamentally a capex story. Hyperscalers and large tech firms are pouring hundreds of billions into GPUs, data centers and networking. Azhar’s AI dashboard work notes that AI capex as a share of GDP and sector revenue is exploding, while AI‑specific revenues are still catching up; in his framing, AI’s industry strain — the ratio of AI revenue to AI capex — sits somewhere in “amber flirting with red,” with multiple dollars of investment for every dollar of current AI revenue.⁴ A mainstream synthesis of this is Fortune’s “AI isn’t a bubble — but it’s showing warning signs.”²¹

Crypto, by contrast, is protocol‑first. Its “capex” is largely R&D, incentives, startup burn and some infra; its revenues are native and visible: CEX and DEX fees, lending spreads, stablecoin income, blockspace fees. The main systemic risk is not “we built too many data centers,” it is “we allowed too much of the revenue stack to depend on one macro regime and one product vertical” — namely, high interest rates making stablecoin reserves extremely profitable and USD rails dominating everything. Stablecoin profit and volume coverage in 2024–2025 — from Tether’s own profit disclosures to “stablecoin payments hit $9T in 2025”‑type pieces and reports that stablecoin volumes now rival or exceed Visa and PayPal — makes this dependence very obvious.⁵ ⁶ ¹² ²²

One‑line version: AI is in danger of a capex bubble; crypto is in danger of a concentration problem.

Escape velocity: how crypto grows 10× from here

Escaping the four‑year halving/ETF/halving/ETF loop requires demand that doesn’t care about what Crypto Twitter thinks on a given day. There are at least three obvious paths.

The first is AI agents as native economic actors. Take the simplest version: a software agent that can reason, hold a wallet, and call APIs. That agent needs to pay for compute, data, other agents, and humans. It needs to manage working capital. It needs to settle millions of tiny transactions at all hours, with no human in the loop. ACH, SWIFT and card networks are not built for that. Public blockchains with programmable money are. Early experiments connecting LLM agents to wallets and on‑chain protocols — the “agentic DeFi” and autonomous agent experiments you see in dev circles — are the pre‑alpha of this world.

The second is machine‑to‑machine payments more broadly. IoT devices paying each other for bandwidth, storage, or live data; microservices paying other microservices for results; bots paying for order flow and execution. Once the average transaction size drops below what card networks can handle economically, but the total count explodes, crypto’s settlement and composability advantages start to dominate. Derivatives and infrastructure reports already show that algorithmic and automated flow is a large share of volume; it’s not a stretch to go from “bots trading” to “agents doing everything.”¹⁷ ¹⁸

The third is boring institutional plumbing. Tokenized treasuries being used as collateral. Cross‑border B2B settlement moving from correspondent banking to stablecoins. Fund subscriptions and redemptions happening on‑chain for auditability and speed. Research from firms like Amundi, Henley, and various bank outlooks all point in this direction: crypto slowly but steadily embedding itself as backend infrastructure rather than front‑of‑house speculation.¹⁴ ¹⁵

None of these rely on another memecoin season. All of them rely on crypto doing what it’s actually good at: being an always‑on, programmable, global settlement layer.

Agents, Relationships, And What Crypto Is Really For

The relationship that will matter most over the next decade may not be human‑to‑human at all. It will be agent‑to‑agent, agent‑to‑protocol, and agent‑to‑marketplace. In that world, “identity” is mostly an on‑chain reputation curve. “Credit” is a function of behavior across thousands of transactions, not a PDF loan application. “Counterparty risk” is quantifiable history, not vibes.

Crypto is oddly well suited to that reality. Histories are public. Contracts are code. Access is permissionless. You can build reputation, credit, coordination, and ownership directly on top of that substrate.

If we still manage to turn that into a bubble, it won’t be because humans got too excited about dog coins a third time. It’ll be because we underestimated how quickly machines would start needing blockspace, liquidity, and programmable money — and how bad humans are at pricing that future.

Citations

1. Economic bubble — Wikipedia
   https://en.wikipedia.org/wiki/Economic_bubble
2. “Understanding the 5 Stages of an Economic Bubble” — Investopedia
   https://www.investopedia.com/articles/stocks/10/5-steps-of-a-bubble.asp
3. “Bubble? 5 for 5.” — Richard Bernstein Advisors (PDF)
   https://assets.ctfassets.net/tpwhn4idgh34/29rN8kzEkFbe5RhcCNF1Hp/3204711951c436b1cb3a25a2f913fee5/RBA_Insights_Bubble_06.21.pdf
4. “Is AI a bubble?” and AI boom/bubble dashboard — Exponential View (Azeem Azhar)
5. “State of Crypto 2025: The year crypto went mainstream” — a16z crypto
   https://a16zcrypto.com/posts/article/state-of-crypto-report-2025/
6. “Stablecoin Payments Hit $9 Trillion in 2025, Rivaling Global Giants Like PayPal and Visa: A16z” — Yahoo Finance
   https://finance.yahoo.com/news/stablecoin-payments-hit-9-trillion-042534119.html
7. “2025 Crypto Adoption and Stablecoin Usage Report” — TRM Labs
   https://www.trmlabs.com/reports-and-whitepapers/2025-crypto-adoption-and-stablecoin-usage-report
8. “The 2025 Global Adoption Index” — Chainalysis
   https://www.chainalysis.com/blog/2025-global-crypto-adoption-index/
9. “DeFi TVL Hits Record $237B as Daily Active Wallets Plunge 22% in Q3” — Yahoo Finance
   https://finance.yahoo.com/news/defi-tvl-hits-record-237b-181957477.html
10. “2025 DeFi Report Highlights Revenue Growth and Value Distribution” — Phemex
    https://phemex.com/news/article/2025-defi-report-highlights-revenue-growth-and-value-distribution-53605
11. “Crypto Protocol Revenue Growth: A Lucrative Opportunity in DeFi and Stablecoin Ecosystems” — AInvest
    https://www.ainvest.com/news/crypto-protocol-revenue-growth-lucrative-opportunity-defi-stablecoin-ecosystems-2508/
12. “Tether Hits $13 Billion Profits for 2024 And All-Time Highs in U.S. Treasury Holdings…” — Tether
    https://tether.io/news/tether-hits-13-billion-profits-for-2024-and-all-time-highs-in-u-s-treasury-holdings-usdt-circulation-and
13. “Crypto & Blockchain Venture Capital — Q4 2025” — Galaxy Digital
    https://www.galaxy.com/insights/research/crypto-blockchain-venture-capital-q4-2025
14. “Cryptocurrencies Break Into The Mainstream 2025” — Amundi
    https://www.research-center.amundi.com/article/cryptocurrencies-break-mainstream
15. “The Crypto Wealth Report 2025” — Henley & Partners
    https://www.henleyglobal.com/publications/crypto-wealth-report-2025
16. “Crypto Bubble 2025: Are We in Another Speculative Frenzy or …” — BTCC
    https://www.btcc.com/en-US/square/NovaFund/887920
17. “CoinGlass 2025 Report: Crypto Derivatives Trends For 2026” — Pi42
    https://pi42.com/blog/crypto-derivatives-market-2025-report/
18. “Perpetual Futures Market Growth and Metrics Analysis in 2025–2026” — KuCoin
    https://www.kucoin.com/news/flash/perpetual-futures-market-growth-and-metrics-analysis-in-2025-2026
19. Bitcoin NVT Ratio — Woobull Charts
    https://charts.woobull.com/bitcoin-nvt-ratio/
20. Bitcoin: Net Unrealized Profit/Loss (NUPL) — Bitcoin Magazine Pro
    https://www.bitcoinmagazinepro.com/charts/relative-unrealized-profit--loss/
21. “AI isn’t a bubble — but it’s showing warning signs” — Fortune
    https://fortune.com/2025/11/13/ai-boom-bubble-bust-dot-com-azeem-azhar/
22. “Stablecoin Monthly Volumes Now Outpace Visa, PayPal As Supply Crosses $300 Billion This Year: Report” — Stocktwits News
    https://www.stocktwits.com/news-articles/markets/cryptocurrency/stablecoin-volumes-now-outpace-visa-paypal-as-supply-crosses-300-billion-this-year-report

Crypto, Bubbles, and The Coming Agent Economy was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.