US July CPI comes in at 3.4% on August 12, 2026 — September hike odds fall to 42%

A day earlier this page reported the consensus for US July CPI as 3.4% headline and 2.5% core, and warned that the figure came from a single forecast aggregator and would differ by house.

The print was 3.4% and 2.5%. Identical to the decimal.

A consensus being right is not the news. The news is what markets did with it: they cut the probability of a September rate hike to 42%.

1. What was released

The most informative cell is the +0.1% monthly. Annual rates are hostage to what happened twelve months ago; the monthly print measures what actually happened in the last four weeks. Annualised, 0.1% is roughly 1.2%. On the most recent month alone, prices were already rising more slowly than the Fed's target.

Shelter at 0.1% matters disproportionately. It carries the largest weight in US CPI, so when shelter is flat the whole index is pinned down with it.

2. The oil question this print did not answer

Yesterday this page flagged that the gap between headline and core had widened to 0.9 points, and that the gap itself encoded an assumption — that energy was pushing prices up. Brent had moved into the mid-$80s per barrel in early August.

The realised gap was also 0.9 points (3.4% minus 2.5%). The structure is unchanged.

But July CPI measures prices paid in July. The August oil move, driven by the unresolved Strait of Hormuz negotiations, has not reached this data yet. So the release splits cleanly:

• Answered — through July, inflation was cooling
• Not answered — whether August energy prices push September's print back up

3. What actually changed was a probability

The wording deserves care, because it inverts the assumption most readers bring. What US markets are pricing is a hike, not a cut. The federal funds target range is 3.50–3.75%, and the Fed has held at that level for five consecutive meetings through July. With inflation stuck in the 3% range, the argument for resuming increases had stayed alive. This print took energy out of that argument.

Equities read it exactly that way.

The Nasdaq led, mirroring the prior session when it fell hardest ahead of the release. The most rate-sensitive assets move first in both directions.

4. Why 3.4% does not feel like relief

Three distinctions clear up most of the confusion between the statistic and the experience.

A falling rate is not falling prices. 3.4% means goods cost 3.4% more than a year ago. For the index level to actually decline, the monthly print would have to be negative. July was +0.1%.

Wages are the comparison that creates the feeling. Wage growth around 3.2% against inflation of 3.4% leaves a 0.2-point shortfall. Small in a month, cumulative over a year. That arithmetic is the honest basis for "inflation is under control but I'm not better off."

An average is not a basket. CPI is a weighted average across hundreds of items. If the things a household buys weekly rise faster than the index, lived experience diverges from the headline regardless of how accurate the headline is.

5. The route into Asian markets

The release landed at 21:30 KST on August 12, after Korean markets had closed. The KOSPI's 3.68% gain that day contains none of this. That move was driven by semiconductor and AI infrastructure news; CPI came afterwards. The inflation effect on Korean flows shows up from the following session.

Two channels run in opposite directions. Lower US hike pressure weakens the dollar and supports emerging-market risk appetite — favourable for foreign flows into Korea. But the stocks that led the August 12 rally trade on earnings expectations more than on rate expectations, which mutes the CPI effect on exactly the names that moved. That session is covered in "KOSPI closes at 6,579.04."

6. What is unresolved

This article does not have the category breakdown. Confirmed items are food and shelter at +0.1% each. How energy, used vehicles, medical care and airfares combined to produce the 0.1% headline was not confirmed. The same 0.1% implies a different September if it came from energy than if it came from services.

42% is not a fixed number. It is backed out of futures pricing and moves substantially within days. It describes the state of the market immediately after the August 12 release and nothing more.

More data arrives before the September meeting. July payrolls already came in weak, and oil has moved the other way since. That divergence — soft labour, firm energy — is the position the Fed is actually in, and one inflation print does not resolve it.