Federal Reserve Beige Book: 10 of 12 districts report growth, and data centres appear in 8 of the 12
The Federal Reserve published its Beige Book on 2 September 2026, prepared at the Minneapolis reserve bank on information collected on or before 24 August. Economic activity increased modestly since early July, with 10 of the 12 districts reporting growth in the slight to moderate range and 2 reporting no change.
That summary is accurate and it is also less interesting than what happens when the previous edition is put beside it.
Fewer districts are growing, but the ones that are grew faster
The mid July edition, prepared at Chicago on information to 6 July, found 11 of 12 districts growing and 1 flat. Two months on the count of growing districts fell by 1, to 10. Read alone that is a deceleration.
The distribution says something different. Districts reporting slight growth, the weakest positive category, fell from 5 to 3. Districts reporting modest growth rose from 4 to 5, and moderate held at 2. So the count at modest or better went from 6 to 7 while the count at slight or flat went from 6 to 5, on our count of the two summaries. The board did not weaken. It hollowed out at the bottom and firmed in the middle, with Philadelphia and St. Louis moving up from slight to modest, Chicago moving down to slight and Kansas City moving down to no change.
Employment points the same way, with a caveat about how it is counted. In July, 5 districts reported modest, moderate or solid gains and 7 reported little to no change. In this edition 3 report modest gains and 4 report slight gains, a total of 7, with 5 unchanged, on our count. That looks like a broadening across 2 districts, but the July residual bucket was “little to no change”, which absorbs slight gains, while this edition breaks slight gains out as a positive category. Part of the move is a change in the report’s own taxonomy rather than in hiring. What is unambiguous is the direction at the top and the bottom: employment rose very slightly overall, no district reported a decline this time, and July had a couple that did. Wage growth was described as modest to moderate in most districts, and the numeric anchors that individual districts published cluster tightly: Philadelphia put merit and cost of living increases at 1 to 3 percent, Atlanta at 2 to 3 percent and St. Louis at about 3 percent.
Price pressure is easing at the top of the range, not across it
Price pressure eased at the top of the range, and by less than the headline adjectives imply. It is worth saying plainly that this is a survey showing inflation pressure abating into a meeting the market expects to deliver a rate increase, which is a tension the report itself does not resolve.
Districts reporting robust price increases halved from 2 to 1, leaving St. Louis alone in that category on the national summary’s classification. Moderate fell from 9 to 8. A modest bucket appeared that did not exist in the July edition, with 2 districts in it. Districts at moderate or above therefore fell from 11 to 9, on our count. The report’s own framing of the direction of travel is narrower still: the pace of price increases was unchanged in 8 districts, slower in 3 and faster in 1, a net of 2 districts decelerating out of 12 on our reading.
Underneath that, energy is doing most of the work. One large retailer in the Cleveland district reported petroleum based product prices up more than 20 percent against 2 to 3 percent for non petroleum products, at least 8 times the midpoint of that range on our calculation, and a manufacturer there described the effect as “shockwaves through the chemical supply chain.” Minneapolis found more than half of firms seeing higher non labour input prices, a substantial majority reporting wholesale prices up 5 percent or more, and nearly a third reporting increases above 10 percent; separately, a builder working on data centre projects there put announced aluminium and steel increases at 5 to 10 percent. Cleveland also recorded a tenth consecutive period of robust non labour cost pressure.
The national outlook paragraph names the sources of uncertainty directly: sentiment was mixed across sectors, “with contacts reporting heightened uncertainty surrounding the effects of higher energy prices, policy, and international conflict.”
Data centres are holding up manufacturing and construction
This is the finding that will outlast the month. Data centres appear as a material source of demand, construction activity or investment pressure in 8 of the 12 district reports, on our count: New York, Cleveland, Richmond, Atlanta, Chicago, St. Louis, Minneapolis and Kansas City. They are not uniformly a positive. Richmond’s contacts raised data centre moratoriums as a reason to expect construction and capital investment to slow.
It shows up on both sides of the ledger. Manufacturing picked up in 9 districts with none declining, on our count, and the national summary attributes that to ongoing strength in demand for defence orders and orders related to data centres. In construction the dependence is more stark. Residential building weakened or declined in 7 districts against 4 rising, on our count, while non residential construction rose on balance, with some districts noting a high concentration of activity tied to data centre projects. A Chicago construction contact put it without qualification: “without data centers, construction would be in a recession.”
The power constraint appears in the same pages. A Kansas City contact observed that “price insensitive technology companies are more concerned about the ‘speed to power’ than the price of power,” which is the demand side of the same transmission and transformer bottleneck that a 25 year capital expenditure forecast for this build, published on the same day, identified as the constraint determining which projects break ground.
Consumer spending was the weakest part of the board, growing slightly on balance with 5 districts up and 4 down on our count, and Cleveland recording a fourth consecutive period of decline. Loan volumes remained solid or increased in most districts, with 7 reporting stronger demand on our count. Chicago is the one district where the two readings diverge: business loan volumes rose slightly there while consumer loan volumes fell modestly and financial conditions tightened slightly.
One note on the filing
The edition published on 2 September is titled Beige Book August 2026 and is filed by the central bank under the August reference. It is the edition that feeds the rate setting meeting of 15 and 16 September, at which the committee also publishes its quarterly projections. On CME Group’s FedWatch measure, traders were pricing about a 66 percent probability of a 25 basis point increase at that meeting when it was read on the morning of 2 September in New York. That gauge is a live reading that moves through the session and is not a settled daily level. The next edition is scheduled for 14 October.
| Pace of activity | 2 Sep | 15 Jul |
|---|---|---|
| Moderate | 2 | 2 |
| Modest | 5 | 4 |
| Slight | 3 | 5 |
| No change | 2 | 1 |
District counts by the adjective each district used to describe overall activity. The counts are ours, taken from the 12 district sections of each edition, and both columns reconcile with the national summaries: 10 growing and 2 flat in September, 11 growing and 1 flat in July. No district reported a decline in either edition.
| Pace of price increases | 2 Sep | 15 Jul |
|---|---|---|
| Robust | 1 | 2 |
| Moderate | 8 | 9 |
| Modest | 2 | 0 |
| Slight | 1 | 1 |
District counts for price growth, as stated in each edition’s national summary. The modest category is new in this edition. Districts at moderate or above fell from 11 to 9, on our count.
| Sector | Districts rising | Districts falling |
|---|---|---|
| Manufacturing | 9 | 0 |
| Loan demand | 7 | 0 |
| Consumer spending | 5 | 4 |
| Residential real estate | 4 | 7 |
Our count of the direction reported in each of the 12 district sections. The balance in each row is flat, mixed or not characterised, so rows do not sum to 12. No district reported an outright decline in overall manufacturing or in overall loan demand, though Chicago’s consumer loan volumes and Kansas City’s residential mortgage demand both fell within districts whose totals held up.
Sources: Federal Reserve, CME Group.
