November 2026 I bond rate: one CPI print decides it (tracker)

The November 2026 I bond rate is not something the Treasury will decide in a room in late October. It is arithmetic, and one of the two numbers that arithmetic needs is already sitting in the Bureau of Labor Statistics database, unchangeable, months in the past. The other number does not exist yet — it will not exist until the September 2026 Consumer Price Index prints in mid-October — and everything genuinely uncertain about November’s rate is compressed into that single missing figure.

Right now, on I bonds issued between May 1 and October 31, 2026, the composite rate is 4.26 percent, built from a 0.90 percent fixed rate and a 1.67 percent semiannual inflation rate, per TreasuryDirect. That composite resets every six months, on May 1 and November 1, and the inflation half of it is the part genuinely in motion this cycle.

Only two CPI numbers decide this rate

The formula the Treasury publishes for the composite rate is: fixed rate plus two times the semiannual inflation rate plus the product of the fixed rate and the semiannual inflation rate. The semiannual inflation rate itself is simpler than most coverage makes it sound — it is the percentage change in the CPI-U, series CUUR0000SA0, between exactly two months, six months apart. For the rate that takes effect in November, those two months are March 2026 and September 2026.

March 2026 CPI-U is already published and fixed at 330.213. It will never change again. September 2026 CPI-U does not exist yet; the Bureau of Labor Statistics publishes it in mid-October. Those are the only two data points the formula touches. April, May, June, July, and August matter only as a trajectory that hints at where September might land — none of them enter the calculation directly. A common misconception treats the I bond rate as some kind of average across the six months of the period. It is not. It is a two-point comparison, and everything between the endpoints is scenery.

Checking the method against the rate already in effect

Before trusting a forecast for November, it is worth confirming the method reproduces a number that is already public. The rate in effect right now — 1.67 percent semiannual, feeding the 4.26 percent composite — was set by comparing September 2025 CPI-U (324.800) against March 2026 CPI-U (330.213):

330.213 ÷ 324.800 − 1 = 1.67%

That is exactly the semiannual inflation rate TreasuryDirect lists for bonds issued May through October 2026. The method is not an approximation of the official calculation; run on the same two endpoints, it produces the official number.

What the data says so far

The Bureau of Labor Statistics has published six of the seven CPI-U readings inside the current window. March, the fixed base, is locked at 330.213. The five months since then are indicative, not determinative:

The last three rows are the detail worth sitting with. The index fell from May’s 335.123 to June’s 333.952 — a real month-over-month decline in the non-seasonally-adjusted series, largely an energy-price effect — and July, published August 12, held essentially flat at 333.918. August broke that streak: the index rose 0.32% to 334.980, published September 11. Only September, the number that actually sets the rate, remains — it arrives in mid-October.

Three scenarios for the September print

None of these are predictions. They are the arithmetic of four assumptions about where September lands, run through the same formula that reproduced the current 4.26 percent rate above, holding the fixed rate at 0.90 percent throughout:

The honest read of that range: outside a further acceleration beyond August’s pace, the November composite is tracking below the 4.26 percent rate paying out now — though August’s rise narrowed the gap considerably. For the composite to actually match today’s 4.26 percent, the semiannual rate would need to reach 1.6725 percent, which requires a September CPI-U of 335.74 or higher — about 0.23% above August’s level. All four scenarios sit on the same assumption flagged above — a 0.90 percent fixed rate carried forward unchanged — which the Treasury has not committed to and does not have to honor.

The October decision

This is the part of the tracker that is actually actionable. Buy an I bond any day through October 31, 2026, and the first six months of that specific bond earn the 4.26 percent composite in effect today, regardless of what November’s rate turns out to be — the new rate only applies starting from the bond’s own six-month anniversary, not to bonds already issued. Buy in November instead, and the bond starts directly on whatever composite is announced then, with no exposure to the 4.26 percent window at all.

Given where the scenarios above point, buying before the deadline still dominates in most of the plausible outcomes: a household locking in October captures six months at 4.26 percent and then rolls into whatever November brings, the same eventual exposure a November buyer gets, plus a better first half-year in every scenario except one. That one branch is no longer a stretch the way it looked over the summer: if September rises 0.23 percent or more over August, to 335.74 or higher, the new composite matches or beats 4.26 percent, and a November buyer starts directly on the richer rate instead of rolling into it from below — the fourth scenario in the table. August’s acceleration brought that outcome meaningfully closer, but it is still not the central case. Because the mid-October CPI release lands before the October 31 deadline, the decision does not have to be made blind; the September number will be public with roughly two weeks still on the clock. Anyone weighing the $10,000 annual purchase limit, or timing a purchase against the three-month early-redemption penalty on an existing bond, should let that mid-October print — not this page’s scenarios — make the final call.

For a fuller comparison of what an I bond actually offers against alternatives with market-priced inflation protection, see I bonds vs. TIPS. And for the mechanics of the vehicle itself — the fixed-rate component, the one-year lockout, the purchase limits — the starting point is I bonds explained.

This tracker uses the same discipline as its sibling piece on the 2027 Social Security COLA: reproduce the government’s own formula against a verified past result, then apply it forward to the unknown inputs. The two calculations are not identical, though. The COLA averages three months of CPI-W against three months from a prior year. The I bond rate compares exactly two months of CPI-U, six months apart, with nothing in between averaged in. Two trackers, two different government formulas, and two different definitions of what counts as “the data.”

This page updates with each CPI-U release through the September report in mid-October, which is the one that actually sets November’s rate.

Sources

• Current I bond composite rate, fixed rate, and semiannual inflation rate, plus the official composite-rate formula: TreasuryDirect — I Bond Rates.
• CPI-U, not seasonally adjusted, U.S. city average, series CUUR0000SA0, underlying the semiannual inflation rate calculation: Bureau of Labor Statistics — CPI databases.

If a CPI reading or rate on this page looks off against the current BLS or TreasuryDirect release, those sources are authoritative; let us know via contact and we will reconcile.