AI’s debt boom is competing with Bitcoin for capital. It may also strengthen Bitcoin’s long-term case

What the AI financing boom changes for Bitcoin - and what it does not prove

1. 011. Five percent is an opportunity cost, not a prediction

   Treasury's official par curve put the 10-year yield at 5.00 percent on 15 September, 5.01 percent on 16 September and 5.17 percent on 25 September. A higher risk-free rate raises the return every risky investment must justify. That can redirect pension funds, income portfolios and leveraged investors toward bonds, but it does not mechanically make Bitcoin fall: adoption, ETF flows, liquidity and expectations can overwhelm the rate effect for long periods.

2. 022. Treasury is buying old bonds for liquidity, not capping the yield

   The video says Treasury tripled long-bond buybacks to support a market under pressure, without success. Treasury's announcement is narrower: from 9 September it increased the maximum for long-end liquidity-support operations from $2 billion to at least $4 billion, at least doubling it. It said the aim was to improve liquidity in older 10-to-30-year securities and cited strong offers from market participants. A buyback exchanges old debt for cash raised through new issuance; it is not Federal Reserve quantitative easing, and Treasury did not say it was trying to hold yields below a target.

3. 033. Alphabet's $84.75 billion was equity, and dilution is part of the price

   Alphabet's SEC filing confirms the upsized raise: public offerings, a planned $40 billion at-the-market programme and a $10 billion private placement with Berkshire Hathaway, for potential gross proceeds of $84.75 billion. The company said the money would expand AI infrastructure and compute. Equity does not promise repayment, but existing owners give up part of their claim on future profits. AI is therefore competing for capital through dilution as well as debt.

4. 044. Amazon showed what long-duration AI financing now costs

   Amazon's 7 July term sheet covers $25 billion across eight tranches: floating-rate notes and fixed-rate maturities from 2029 to 2066, with fixed coupons from 4.6 to 6.25 percent. That is not evidence Amazon was unable to borrow; the notes were rated in the A1 to AA range. It is evidence that investors can obtain contractual income from a highly rated company while gaining indirect exposure to the same AI buildout that supports the equity story.

5. 055. AI debt can add pressure, but it did not single-handedly create 5 percent Treasuries

   The BIS has documented hyperscalers' growing use of debt and private credit to finance AI capital expenditure, including tighter links with insurers and other non-bank investors. More long-duration corporate supply can require higher yields to clear and can compete with Treasury issuance for buyers. But September also brought elevated inflation, geopolitical energy shocks and a Federal Reserve rate increase to 3.75–4 percent. The evidence supports AI financing as one source of bond supply, not a clean causal claim that data centres pushed the 10-year above 5 percent.

6. 066. The short-term Bitcoin mechanism is supported by research

   An IMF working paper found that one common crypto factor explained most crypto-price variation in its sample, that crypto's correlation with equities rose alongside institutional participation, and that Federal Reserve tightening reduced that factor through the same risk-taking channel seen in stocks. In plain language: when safe yields rise and financial conditions tighten, Bitcoin is still commonly priced as a risky asset. The paper is empirical evidence about past behaviour, not a rule for every market episode.

7. 077. Issuer-free is a real distinction, not yet a reserve-asset verdict

   A Treasury is a claim on the US government, a corporate bond is a promise from a company and a share is a residual claim on its profits. Bitcoin is different: holding it directly is not a claim on an issuer's balance sheet. That removes default by a named issuer, but it does not remove custody risk, market risk, protocol risk or the chance of severe drawdowns. Scarcity and independence describe the asset; reserve status describes how institutions actually use and behave around it.

8. 088. Gold shows the destination, not a shortcut

   The World Gold Council's 2026 survey found that 45 percent of 76 responding central-bank reserve managers expected to increase their own gold holdings in the next year, with crisis performance, store of value and diversification among the leading reasons. That supports demand for assets outside another issuer's promise. It does not establish that Bitcoin is “gold but better”: gold has deep official markets, a much longer price history and far lower drawdown risk. Bitcoin must earn the comparison through behaviour, not supply mathematics alone.

9. 099. The Czech central-bank example is a test, not reserve adoption

   The Czech National Bank did buy a $1 million portfolio containing Bitcoin, dollar stablecoins and a tokenised deposit in November 2025. Its own statement is explicit that the portfolio sits outside international reserves, represents 0.0006 percent of assets and will not receive more money during the two-to-three-year test. A later study found that a hypothetical 1 percent allocation improved historical risk-return results, while warning that Bitcoin's correlations are unstable and its short history makes those results unreliable as a forecast.

10. 1010. Bank rules still treat unhedged Bitcoin as exceptionally risky

    The Basel framework applies a 1,250 percent risk weight to Group 2b crypto exposures, designed to require capital at least equal to the exposure, and generally limits aggregate Group 2 exposure relative to Tier 1 capital. The committee is conducting an expedited review, and implementation differs by jurisdiction, so “the rule applies to every bank from 1 January 2026” is too broad. The direction is still clear: under the current international standard, Bitcoin is nowhere near the prudential treatment of sovereign reserves.

11. 1111. The decisive test is a crisis of trust, not another bull market

    Bitcoin can rise while rates are high and still remain a risk asset. A stronger reserve case would require lower structural volatility, less dependence on leverage and arbitrage holders, workable custody and bank rules, and repeated evidence that long-horizon holders do not sell into stress. Most importantly, it would need to hold value during a crisis of confidence in a major issuer without simply falling alongside technology shares in a liquidity rush. That test has not been passed consistently.

Where we track this

• Is the AI boom a bubble? →

  A separate look at what is real, what is financed and where the buildout becomes fragile.

• AI's bottleneck is increasingly energy →

  Why data centres, grid connections and power infrastructure turn software demand into long-duration capital spending.

• Why cheaper AI can still raise the total bill →

  The demand mechanism that can keep infrastructure spending high even as the price of one inference falls.

Questions people ask

Does AI debt make Bitcoin go down?

Not mechanically. More attractive government and corporate yields increase Bitcoin's opportunity cost and tighter financial conditions have historically weighed on crypto through the risk-taking channel. Bitcoin can still rise because of adoption, ETF flows, liquidity or changing expectations. AI debt is a headwind in the framework, not a price forecast.

Is the US Treasury buying back debt to stop yields rising?

Treasury says its buybacks support liquidity in older securities, not a target yield. From 9 September 2026 it raised the maximum size of long-end operations from $2 billion to at least $4 billion. These operations exchange older debt while Treasury continues issuing new debt; they are not the same as Federal Reserve quantitative easing.

Why would AI companies issue debt if they have large cash flows?

Data centres, chips and power infrastructure require enormous spending before the related revenue arrives. Debt and equity let companies spread that financing across investors and maturities instead of funding every project from current cash flow. The strategy works when the return on the new capacity exceeds its financing cost; that outcome is not guaranteed.

Is Bitcoin already a reserve asset like gold?

Not in institutional behaviour or regulation. Bitcoin has issuer-free scarcity, but it remains highly volatile, often correlates with equities during stress and receives punitive treatment under the current Basel bank-capital framework. The Czech central bank's $1 million portfolio is explicitly a technology test outside its international reserves, not reserve adoption.

What would make the reserve-asset case for Bitcoin stronger?

Lower and more stable volatility, deeper unleveraged ownership, mature custody and accounting rules, more workable prudential treatment and repeated evidence that Bitcoin holds value during crises of confidence in major issuers - not only during crypto bull markets - would all strengthen the case.