Crypto, by contrast, has looked weaker on the surface. U.S. spot Bitcoin ETFs built an important institutional access layer after launch, but 2026 flows have been uneven, with multiple sources showing sustained May and June outflows and SoSoValue snapshots putting cumulative Bitcoin ETF net inflows near $53.85 billion in early June after earlier, higher snapshots were reduced by the outflow wave ([6], [7]). Ethereum ETF data has been similarly uneven, with SoSoValue-sourced reports showing repeated weekly outflows into June 2026 despite cumulative net inflows remaining positive since launch ([8]).

The more interesting development has been beneath the surface. AI took the market’s attention, while crypto rebuilt the plumbing. The next crypto cycle is less likely to resemble the retail-reflexive cycle of 2021 and more likely to be infrastructure-led: regulated stablecoin frameworks, ETF access, qualified custody, tokenized asset rails, and programmable settlement that may become increasingly relevant to an AI-native internet. Stablecoins are the clearest evidence of that shift, but they also require definitional discipline: BIS cited annual stablecoin transaction volume of roughly $28 trillion in 2025, while also noting that genuine payment activity after deeper filtering is closer to $390 billion, meaning headline transfer volume and economic payment volume are very different concepts ([9]).

This is not a call that one technology wins and the other loses. AI can remain a major technological shift even if the equity trade becomes crowded and harder to extend. Crypto can look weak in price and flows while its underlying rails continue to improve.

The rotation we are watching is simpler: when too much good news is already priced into one theme, capital may begin to look for the infrastructure that has been built quietly elsewhere.

Yes, AI Stole the Show

AI did not simply become a market theme. It became the dominant expression of risk appetite.

Nvidia became the public-market benchmark for the AI cycle. FY2025 revenue reached $130.5 billion, up 114% year over year, while Q4 FY2025 data-center revenue reached $35.6 billion, up 93% year over year and roughly 91% of total revenue ([1]). Q1 FY2026 revenue then rose to $44.1 billion, with data-center revenue of $39.1 billion, up 73% year over year despite a $4.5 billion H20 export-control charge ([10]). Nvidia’s market capitalization moved from roughly $360 billion in January 2023 to more than $4 trillion by July 2025 and above $5 trillion by June 2026, according to Macrotrends’ market-cap series ([11]).

The leadership was not isolated to one stock. The Magnificent 7 delivered an average total return of 60.5% in 2024, while the S&P 500 returned roughly 25%, and their combined S&P 500 weight rose from 21.9% in 2020 to more than 30% by the end of 2024 ([2], [3]). The semiconductor cycle also broadened, with WSTS estimates cited in Nasdaq index research showing global semiconductor revenue rising from $631 billion in 2024 to an estimated $772 billion in 2025 and $975 billion in 2026 ([12]).

Private capital followed the same pattern. PitchBook data cited by TechCrunch put global generative-AI venture funding at $56 billion across 885 deals in 2024, up 92% from 2023 ([4]). Stanford HAI’s 2025 AI Index reported $109.1 billion of U.S. private AI investment in 2024 and $252.3 billion of global corporate AI investment across all forms, including M&A, private investment, minority stakes, and public offerings ([13]). Crunchbase reported that AI-related companies had raised $118 billion through August 15, 2025, already exceeding its full-year 2024 AI-related total of roughly $108 billion, with eight companies accounting for $73 billion of that 2025 total ([5]).

Capex turned the AI narrative into a balance-sheet cycle. Platformonomics estimated that the four major hyperscalers spent more than $416 billion of calendar-year 2025 capex, up 66% from roughly $251 billion in 2024, with 2026 guidance pointing above $600 billion ([14]). Goldman Sachs’ “Tracking Trillions” framework puts the potential five-year AI capital investment range at $4 trillion to $8 trillion depending on assumptions about chip efficiency, refresh cycles, power cost, and end-demand ([15]).

Crypto’s scoreboard looked different. Galaxy reported that Q1 2026 crypto and blockchain venture investment totaled $4.0 billion across 355 deals, down 50% quarter over quarter from Q4 2025, though still above many 2023 to 2024 trough-quarter readings ([16]). U.S. spot Bitcoin ETFs remain a major structural achievement, but 2026 flows were uneven, including a 13-session outflow streak from mid-May to early June and a record weekly outflow around the week of June 2 to 6 in multiple SoSoValue and Farside-referenced reports ([17], [18]).

It is clear that AI owned the cycle, on the other hand, crypto’s work was quieter: repairing the damage from 2022 and rebuilding the rails that could make the next cycle more institutional than speculative.

The Fragility of a Crowded Trade

The fragility of the AI trade does not come from skepticism about the technology. It comes from the gap that can open between technological transformation and financial-market pricing.

Goldman Sachs framed this tension clearly in “Gen AI: Too Much Spend, Too Little Benefit?”, asking whether roughly $1 trillion of coming AI capex could be justified by end-market revenues and featuring Daron Acemoglu’s estimate that AI may add only about 0.5% to U.S. productivity over the next decade under current deployment assumptions ([19]). Sequoia’s “AI’s $600B Question” used infrastructure run-rate economics to estimate that the AI ecosystem needed roughly $600 billion of annual revenue to justify the buildout implied by Nvidia’s revenue trajectory and downstream gross-margin requirements ([20]). JPMorgan Asset Management framed a similar revenue gap, estimating that a 10% return on AI investments would require roughly $650 billion of annual AI revenue versus roughly $25 billion of actual industry revenue at the time of its analysis ([21]).

AI productivity gains can be real while AI-linked assets still price too much of the future too early. Goldman Sachs later argued that the sector was “not in a bubble, at least not yet,” but also acknowledged rising valuations, massive spending, and circularity as real concerns ([22]). Bloomberg’s work on circular AI deals documents the structure behind those concerns: cloud providers invest in AI labs, AI labs spend heavily on the same providers’ compute, and reported AI demand becomes intertwined with financing architecture ([23]).

Positioning risk is visible as well. BofA’s October 2025 Global Fund Manager Survey, reported by Bloomberg and Investing.com, found that 54% of surveyed investors believed AI stocks were in a bubble and that “Long Magnificent 7” was the most crowded reported trade ([24], [25]). Fortune later reported that BofA’s November 2025 survey showed 45% of institutional investors naming “AI bubble” as the biggest tail risk and, for the first time in more than 20 years, fund managers saying companies were overinvesting ([26]).

An unwind in enthusiasm would not require AI to disappoint in absolute terms. It would only require the rate of positive surprise to slow. Once the market has priced extraordinary adoption, extraordinary capex, extraordinary margins, and extraordinary equity concentration, the burden of proof shifts from possibility to monetization. In that environment, a still-transformative technology can become a more difficult financial trade.

Crypto’s Quiet Rebuild

Crypto’s weaker narrative intensity should not be confused with the absence of structural progress.

Access was the first step. U.S. spot Bitcoin ETFs turned Bitcoin into a security-wrapper asset that traditional allocators could hold through familiar operational infrastructure. SoSoValue data cited in early June placed cumulative Bitcoin ETF net inflows near $53.85 billion after the outflow wave, while earlier Farside-referenced snapshots had shown higher cumulative figures before May and June redemptions ([6], [27]). The near-term flow picture is mixed: institutional access has been established, but 2026 demand has not been linear.

This updates our April 2026 work on the Bitcoin spot ETF structural-bid thesis and Morgan Stanley’s MSBT launch. The ETF wrapper created institutional access, but 2026 has made the distinction between access and persistent marginal demand more visible. The structural bid remains relevant, yet it has become more flow-sensitive, more allocator-driven, and more exposed to competing AI-led risk appetite.

Implementation also improved. Custody, prime brokerage, and regulated trust-bank infrastructure have continued to mature, with multiple digital asset firms pursuing or receiving OCC national trust bank approvals and with the OCC clarifying non-fiduciary custody authority for national trust banks in 2026 ([28]). These developments matter less for headlines than for allocation plumbing. Large pools of capital cannot treat digital assets as operationally mature unless custody, reporting, control, and qualified-custodian frameworks meet institutional requirements.

Venture activity has become more selective. Galaxy reported $4.0 billion invested in crypto and blockchain startups across 355 deals in Q1 2026, with the annualized run rate below 2025’s roughly $20 billion but still above much of the 2023 to 2024 downturn pace ([16]). The composition also looks more infrastructure-heavy than the 2021 cycle, with trading, exchange, investing, lending, infrastructure, payments, tokenization, and DeFi among the reported categories ([16]).

Regulation has become more concrete. The GENIUS Act created the first federal U.S. stablecoin framework, with the White House describing requirements for 100% reserve backing, monthly public reserve disclosures, and strict marketing rules around federal backing and legal-tender claims ([29]). FinCEN and OFAC have also moved to implement AML and sanctions compliance for permitted payment stablecoin issuers, treating them as financial institutions under the Bank Secrecy Act and requiring risk-based sanctions compliance programs ([30]).

None of this guarantees better price action. It does suggest that crypto is returning with a different institutional architecture than it had during the last retail-led cycle.

Stablecoins: The Strongest Evidence of Product-Market Fit

Stablecoins are the strongest utility evidence in crypto because they have moved beyond abstract optionality. They are already used as digital-dollar rails, trading settlement instruments, cross-border transfer tools, and increasingly as programmable payment infrastructure. The difficulty is that “stablecoin volume” is one of the most abused phrases in digital-asset research.

The broadest figures are impressive but not sufficient. BIS cited annual stablecoin transaction volume of roughly $28 trillion in 2025, but it also emphasized that this was less than three business weeks of settlement volume for the largest U.S. wholesale payment systems and that genuine adjusted payment values were closer to $390 billion annually ([9]). a16z crypto reported stablecoin transaction volume of roughly $46 trillion over the prior year and adjusted transaction volume near $9 trillion, while also emphasizing that adjusted methodologies attempt to filter artificial activity and bot-driven flows ([31], [32]). Crossmint, citing Allium, reported $11.6 trillion of adjusted transfer volume in 2025 and $374.5 billion of labeled payment volume across 1.1 billion transactions ([33]).

For us , the correct interpretation is a hierarchy. Raw on-chain volume measures token movement. Adjusted transfer volume attempts to remove bots, MEV, wash trading, and obvious artificial patterns. Labeled payment volume attempts to identify actual economic exchange between parties. The headline number teaches us that stablecoins are liquid and deeply embedded in crypto market structure. The narrower number teaches us that real-world payments remain early, but no longer theoretical.

Supply growth supports the same conclusion. BIS put total stablecoin market capitalization near $320 billion at end-May 2026, while StablecoinBeat showed a live snapshot near $305.9 billion on June 23, 2026 after a pullback from the May high ([9], [34]). BIS also noted that 99.4% of fiat-backed stablecoins by market value were USD-pegged, which reinforces the interpretation of stablecoins as digital-dollar distribution rails rather than a general basket of global digital money ([9]).

The GENIUS Act strengthens the institutional case. The White House described the law as creating a federal regulatory system for stablecoins, including 100% reserve backing with liquid assets such as dollars and short-term Treasuries, monthly reserve disclosures, and requirements designed to prevent claims of federal backing or legal-tender status ([29]). Treasury’s implementation work through FinCEN and OFAC links stablecoin issuance to BSA, AML, and sanctions compliance, which is central if stablecoins are to move from crypto-native settlement into regulated commercial payment flows ([30]).

The implication is material: the most important part of crypto may no longer be speculation. It may be monetary infrastructure: dollar-denominated, programmable, globally available settlement rails that can serve both human and machine participants.

The AI-Crypto Convergence

AI and crypto are often framed as competitors for marginal capital. In the short run, that framing is fair. Structurally, the relationship may become indispensably complementary,  crypto is built for AI. 

AI agents will need wallets, payments, identity, authentication, and audit trails if they are to transact autonomously. Coinbase AgentKit explicitly frames this as “every agent deserves a wallet” and provides infrastructure for AI agents to hold, send, receive, and use crypto assets programmatically ([35], [36]). The x402 protocol, developed around the HTTP 402 “Payment Required” standard, is designed to let autonomous agents pay for online resources through stablecoin payments and related machine-native settlement flows ([37]). Cloudflare’s agentic payments documentation describes AI agents discovering, paying for, and consuming resources programmatically through 402-based payment flows and related protocols ([38]).

Stablecoins will become the natural payment rail for machine-to-machine and agent-to-agent transactions. The use case is not yet macro-scale, but the design logic is clear: agents need low-friction, API-native, global settlement, and stablecoins already provide programmable dollar units on public networks. The early infrastructure is live; the scale is coming.

The Liquidity Lens

Bitcoin has historically behaved as a high-beta expression of global liquidity, but the relationship is not mechanical.

Lyn Alden and Sam Callahan found that Bitcoin moved in the same direction as global M2 in 83% of rolling 12-month periods from May 2013 to July 2024, with a full-period price-level correlation of 0.94 and weaker correlations over shorter windows ([44]). CF Benchmarks’ work similarly frames Bitcoin as highly sensitive to M2, estimating Bitcoin’s M2 beta near 11.3 versus roughly 1.3 for gold and roughly 1.1 to 1.5 for global equities ([45]).

The recent decoupling matters precisely because the historical relationship has been strong. CF Benchmarks argues that global M2 grew more than 12% over the 12 months to February 2026 while Bitcoin declined roughly 12%, creating a roughly 24 percentage-point divergence and pushing Bitcoin’s model-based Z-score from +1.48 in January 2025 to -1.31 in February 2026 ([45]). That gap supports a monitoring framework, not a forecast.

This extends prior Axys previous memo that we published last month. In that framework, Global M2 momentum, including ROC(13) forwarded by roughly 13 weeks, is treated as an indicator of cycle pressure rather than a trade trigger. The present memo applies the same thesis : liquidity may be improving beneath the surface, but ETF flows, dollar strength, real yields, and AI-led risk appetite determine whether that liquidity reaches digital assets.

We view global liquidity as a long-run gravitational force rather than a timing tool. The more relevant question is whether liquidity expansion is beginning to transmit back into digital assets after being interrupted by dollar strength, rate uncertainty, ETF outflows, and competing AI-led risk appetite. DXY, real yields, Fed path, ETF net flows, and the breadth of risk appetite can override liquidity in the short run. Liquidity is best treated as a monitoring overlay: useful for context, dangerous as a standalone prediction engine.

Conclusion 

Crypto does not need AI to fail. Its stronger case is that AI makes crypto more necessary. While AI captured attention through earnings, chips, and capex, crypto rebuilt the rails that an autonomous digital economy may require: ETFs, custody, stablecoins, tokenized assets, programmable settlement, and on-chain auditability. That is the more important setup. 

If AI agents are expected to act, transact, authenticate, and settle on their own, they will need financial infrastructure that is global, programmable, verifiable, and available outside traditional banking hours. Stablecoins and public blockchains are not the only possible answer, but they are the most developed version of that architecture today. This is where crypto’s role becomes more essential. A machine-native internet will need payments, wallets, identity, authentication, provenance, audit trails, and atomic settlement. Crypto is the financial structure most naturally built for that environment. If AI becomes the intelligence layer of the internet, crypto may become the monetary and settlement layer beneath it.

Disclaimer

This material is provided for information purposes only and does not constitute investment advice, a recommendation, or an offer or solicitation to buy or sell any financial product or service in any jurisdiction.

The views expressed are those of the author as at the date of publication and are subject to change without notice. While reasonable care has been taken, no representation or warranty, express or implied, is made as to the accuracy, completeness, or reliability of the information contained herein, and no reliance should be placed upon it.

This material has not been reviewed or approved by the Dubai Financial Services Authority, the British Virgin Islands Financial Services Commission, or the Cayman Islands Monetary Authority, or any other regulatory authority.

This material is not intended for distribution to, or use by, any person in any jurisdiction where such distribution or use would be contrary to local law or regulation. In particular, this material does not constitute a public offer of securities in the British Virgin Islands or the Cayman Islands.

Any investment or service referred to herein will be available only to persons who meet the relevant eligibility criteria under applicable laws and regulations, including, where applicable, Professional Clients (as defined by the Dubai Financial Services Authority) or other equivalent categories of sophisticated or professional investors under applicable law.

Investments involve risk, including the possible loss of capital. Past performance is not a reliable indicator of future results.

“Axys” is a trading name used by a network of independent businesses. Axys does not operate as, and is not, a separate legal entity in any jurisdiction and has no distinct legal personality. Each business operating under the Axys name is independently owned and operated.

Axys Capital  Ltd is authorised and regulated by the Dubai Financial Services Authority. Axys Investment Management Ltd is authorised and regulated by the British Virgin Islands Financial Services Commission.

References

1. Nvidia results via Nasdaq: https://www.nasdaq.com/articles/nvidia-reports-record-q4-revenue-393-billion-full-year-revenue-1305-billion-fiscal-2025

2. Man Group: https://www.man.com/insights/views-from-the-floor-2025-january-14

3. First Trust Portfolios: https://www.ftportfolios.com/Commentary/EconomicResearch/2025/1/8/the-sp-500-index-in-2024-a-market-driven-once-again-by-the-magnificent-seven

4. TechCrunch: https://techcrunch.com/2025/01/03/generative-ai-funding-reached-new-heights-in-2024/

5. Crunchbase: https://news.crunchbase.com/venture/big-dollar-ai-investors-2025-softbank/

6. WEEX citing SoSoValue: https://www.weex.com/news/detail/bitcoin-spot-etfs-see-9137-million-in-daily-net-outflows-led-by-blackrocks-ibit-at-233-million-pu77tup2i8k5j4l5oudmdrn7

7. TFTC: https://www.tftc.io/bitcoin-etf-outflows-2026-record-streaks/

8. WEEX citing SoSoValue: https://www.weex.com/news/detail/data-ethereum-spot-etf-had-a-net-outflow-of-149072-million-last-week-marking-five-consecutive-weeks-of-net-outflows-vxfrog1qoicbpb0i2r5gjoum

9. BIS Annual Economic Report 2026, Chapter III: https://www.bis.org/publ/arpdf/ar2026e3.htm

10. Nvidia: https://nvidianews.nvidia.com/news/nvidia-announces-financial-results-for-first-quarter-fiscal-2026

11. Macrotrends: https://www.macrotrends.net/stocks/charts/NVDA/nvidia/market-cap

12. Nasdaq Global Indexes: https://indexes.nasdaqomx.com/docs/202601 Semiconductor Research - SOX.pdf

13. Stanford HAI: https://hai.stanford.edu/ai-index/2025-ai-index-report/economy

14. Platformonomics: https://platformonomics.com/2026/02/follow-the-capex-2025-retrospective/

15. Goldman Sachs: https://www.goldmansachs.com/insights/articles/tracking-trillions-the-assumptions-shaping-scale-of-the-ai-build-out

16. Galaxy Research: https://www.galaxy.com/insights/research/crypto-blockchain-vc-venture-capital-startups-fundraising-q1-2026

17. Gate.com citing Galaxy Research: https://www.gate.com/blog/btc-etfs-see-13-consecutive-days-of-outflows-with-3-4-billion-in-weekly-net-redemptions-why-are-institutions-selling-btc

18. Investing.com: https://www.investing.com/analysis/bitcoins-34-billion-etf-bleed-looks-more-cyclical-than-structural-200681474

19. Goldman Sachs: https://www.goldmansachs.com/insights/top-of-mind/gen-ai-too-much-spend-too-little-benefit

20. Sequoia Capital: https://sequoiacap.com/article/ais-600b-question/

21. JPMorgan Asset Management: https://am.jpmorgan.com/us/en/asset-management/liq/insights/market-insights/market-updates/on-the-minds-of-investors/how-is-ai-being-monetized.html

22. Goldman Sachs: https://www.goldmansachs.com/insights/goldman-sachs-exchanges/are-ai-bubble-concerns-warranted-or-overblown

23. Bloomberg: https://www.bloomberg.com/graphics/2026-ai-circular-deals/

24. Investing.com: https://www.investing.com/news/stock-market-news/bofas-survey-shows-54-of-investors-say-ai-in-bubble

25. Bloomberg: https://www.bloomberg.com/news/articles/2025-10-14/most-investors-say-ai-stocks-are-in-a-bubble-bofa

26. Fortune: https://fortune.com/2025/11/18/fund-managers-fear-ai-bubble-say-companies-overinvested-bofa-survey/

27. TradingNews: https://www.tradingnews.com/news/bitcoin-etf-inflows-ibit-etf-at-43-usd-with-65b-usd-flows

28. TheFinanceWall: https://www.thefinancewall.com/cryptocurrency/occ-charter-wave-how-circle-coinbase-and-fidelity-digital-assets-are-reshaping-crypto-custody/

29. White House: https://www.whitehouse.gov/fact-sheets/2025/07/fact-sheet-president-donald-j-trump-signs-genius-act-into-law/

30. U.S. Treasury: https://home.treasury.gov/news/press-releases/sb0435

31. a16z crypto: https://a16zcrypto.com/posts/article/big-ideas-things-excited-about-crypto-2026/

32. a16z crypto stablecoin charts: https://a16zcrypto.com/posts/article/stablecoin-data-charts/

33. Crossmint: https://www.crossmint.com/learn/stablecoin-industry-trends-2026

34. StablecoinBeat: https://stablecoinbeat.com/tracker/

35. Coinbase AgentKit: https://www.coinbase.com/developer-platform/products/agentkit

36. GitHub: https://github.com/coinbase/agentkit

37. Coinbase x402: https://www.coinbase.com/developer-platform/products/x402

38. Cloudflare: https://developers.cloudflare.com/agents/tools/payments/

39. C2PA: https://c2pa.ai/news/five-years

40. C2PA: https://c2pa.org

41. ChainScore Labs: https://www.chainscorelabs.com/en/blog/ai-x-crypto-agents-compute-and-provenance/verifiable-computation-and-zkml/why-zk-proofs-are-the-missing-link-for-ai-x-crypto

42. Akash Network: https://akash.network/blog/akash-2025-year-in-review/

43. ForkLog: https://forklog.com/en/depin-after-the-boom-revenue-is-king-as-energy-and-ai-set-the-pace/

44. Lyn Alden: https://www.lynalden.com/bitcoin-a-global-liquidity-barometer/

45. CF Benchmarks: https://www.cfbenchmarks.com/blog/the-m2-bitcoin-relationship-what-the-data-actually-shows