The Money Squeeze Is Real, but Rate Hikes Cause Inflation Is Too Simple

The money squeeze is real: long-term borrowing costs rose sharply while several central banks tightened policy in the same month. But the stronger claim in Casual Finance’s The Money Squeeze That’s Choking the Economy—that higher rates may be adding to inflation because AI investment will not slow—runs ahead of the evidence.

Method note: YouTube exposed the video’s identity and published outline, but its caption endpoint did not return a usable transcript during this review. This is therefore a source-by-source check of the public argument, not a claim that every spoken sentence was transcribed. I also excluded the attachment’s exact Bill Ackman quotation because I could not independently recover the original post, and I dropped its $740 billion AI-capex estimate because the total changes with the companies and spending categories included.

Fact-check scoreboard#

The seven-bank wave happened, but the dates are not equivalent#

The BIS daily policy-rate data show New Zealand moving from 2.50% to 2.75% on September 3; the euro area from 2.25% to 2.50% on September 16; the U.S. midpoint from 3.625% to 3.875% on September 17; Japan from 1.00% to 1.25% on September 24; and Norway and South Africa moving to 4.50% and 7.25% on September 25. These are dates when rates appear in the daily series, not necessarily the meeting dates in the video.

Australia was the final case. The Reserve Bank of Australia’s September 29 statement says its board unanimously raised the cash-rate target by 25 basis points to 4.60%. Calling this a coordinated campaign would be too strong: each bank cited its own inflation and capacity risks. Still, seven hikes inside four weeks are a genuine global tightening cluster.

The bond market supports the real-rate story, not falling breakevens#

The cleanest part of the video’s thesis is the repricing of real money. The Treasury’s daily nominal yield series and daily real yield series let us compare like dates. On August 31, the nominal 10-year yield was 4.75% and the real yield was 2.44%. On September 30 they were 5.29% and 2.93%. That is a 54-basis-point nominal rise, 49 points of which came from the real yield.

That decomposition matters. It says investors demanded much more return after expected inflation, consistent with tighter financial conditions, heavier competition for capital, or a larger term premium. It does not say inflation compensation collapsed. Nominal minus real was 2.31% on August 31, 2.36% on September 30, and also 2.36% when the nominal yield reached 5.31% on October 5. This subtraction is only a proxy because nominal Treasuries and TIPS have different liquidity and risk characteristics, but it directly contradicts a broad claim that breakevens fell across the episode.

The auction evidence is equally stark. A filtered Treasury FiscalData query returns the September 17 auction at a 2.653% high yield. The only other result meeting that threshold back to October 2008 is the October 8, 2008 auction at 2.850%. For readers who want to reproduce government-data queries, Orthogonal’s Treasury FiscalData API guide explains the same no-key API family.

SoftBank proves expensive financing was available, not that demand is immune#

The SoftBank example is accurate after correcting one number. Latham & Watkins’ transaction summary describes US$10 billion of dollar notes in three tranches plus €1 billion in two tranches, about US$11.1 billion in total. It calls the sale the largest global high-yield corporate bond offering and says proceeds would fund the final US$10 billion tranche of SoftBank’s US$30 billion follow-on OpenAI investment, with the balance for general purposes. The same page gives US$14.6 billion as SoftBank’s total high-yield issuance during 2026. That is likely the figure the attachment accidentally treated as one deal.

What does this establish? One strategically committed borrower accepted costly financing and found buyers. It does not establish that aggregate investment is insensitive to rates. SoftBank may view the OpenAI commitment as non-negotiable, while thousands of lower-return data-center, housing, factory, and small-business projects are delayed or cancelled. Observing the project that cleared the market tells us nothing about projects that disappeared before issuance.

Higher rates can raise some costs without becoming inflationary overall#

The video’s mechanism is possible at the firm level. A company refinancing debt may try to pass higher interest expense into prices. Housing developers facing dearer construction loans may require higher selling prices, and utilities or infrastructure providers may seek regulated increases. If demand is hard to replace and supply is constrained, some cost pass-through can survive.

But this is only one channel. Higher rates also lower asset values, weaken housing demand, make inventories dearer to hold, reduce credit creation, and raise the return hurdle for new investment. Firms cannot automatically pass financing costs to customers; competitors, contracts, and weak demand can force them to accept lower margins instead. The net price effect depends on market power, debt maturity, supply constraints, and how much spending is actually abandoned.

AI makes transmission uneven, not absent. Large technology companies and a committed investor such as SoftBank can keep spending while households, builders, and smaller firms retreat. Orthogonal’s Muse, Amazon, and META thesis covers why platform owners may treat AI distribution as strategic. That strategic motive does not turn every AI project into rate-insensitive demand.

September CPI cannot grade September rate hikes#

The CPI figures in the attachment are correct. CNBC’s report of the BLS release records a 0.6% monthly and 3.7% annual headline increase, with core CPI up 0.3% for the month and 2.6% for the year. It also reports that energy rose 5.2% in September and gasoline rose 9.1%, accounting for about half of the headline monthly increase. That makes the release evidence of an inflation problem, but the composition is important: a large energy shock is not proof of demand created by a recent rate decision.

The causal interpretation fails on timing. September CPI measures prices during September. The Fed decision came on September 16, and several other rate changes took effect even later. The report was released on October 14. It can show that inflation pressure remained; it cannot show that decisions made days earlier caused the month’s inflation, nor can one monthly print establish that tighter policy will never reduce demand.

My view: this is a two-speed transmission problem#

The useful conclusion is narrower than “rate hikes cause inflation.” Real capital became much more expensive, yet strategic AI spending and government borrowing did not instantly stop. Monetary restraint may therefore hit housing, small firms, and marginal investment before it materially slows hyperscaler infrastructure or fiscal borrowing. That distribution is economically painful and politically important.

The policy-error test should be forward-looking. Watch whether private credit volume and housing activity contract, whether AI debt issuance broadens beyond a few strong sponsors, whether profit margins absorb interest costs or consumer prices do, and whether the nominal-minus-real yield gap rises with inflation expectations. If strategic spending stays firm while rate-sensitive supply investment collapses, the squeeze can damage capacity before inflation falls. If aggregate credit and demand weaken over subsequent quarters, the standard transmission mechanism is working—unevenly, but not backwards.