Saudi Construction Cost Index: GASTAT CCI 103.99, July 2026

The Saudi construction cost index stood at 103.99 in July 2026 on a 2023 = 100 base, up 2.26% on July 2025 and only 3.99% above that base three years on (GASTAT, released 18 August 2026). The fastest section is equipment and machinery rental, 19.01% of the basket and up 4.53%, against 1.80% for materials and 1.51% for labour. Put GASTAT's with-operator and without-operator hire rates side by side and that plant-hire number is largely a wage, which on our reading puts 43.55% of the index on prices that contain one. Two corrections before anyone quotes it: it is an input-cost index for buildings only, and the section weights on the workbook's front sheet are not the index's weights.

Every figure below comes from a workbook, metadata page or portal we opened ourselves, and the page sits inside our map of Saudi public property data.

Key figures at a glance

What the Saudi construction cost index measures, and the weights people misread

Four input blocks on a basket fixed in 2023: basic materials 48.52% of the index, labour 29.93%, equipment and machinery rental 19.01%, energy 2.54%, printed to fifteen digits in Table 2 of the July 2026 workbook and summing to 100.000000 (GASTAT). Residential building is 77.45% of it, non-residential 22.55%, and the two exhaust it.

That list is also what is missing: no land, no design fee, no preliminaries, no overhead or profit, no finance cost, nothing that converts to a riyal per square metre. Coverage is buildings (GASTAT CCI FAQ), so roads, utilities and the earthworks under a giga-project sit outside it entirely.

The misreading that costs professionals most is one sheet's doing. Every workbook opens on a Summary sheet carrying a second set of section weights: materials 37.526669%, labour 23.045405%, equipment rental 14.924068%, energy 1.955175% (GASTAT, July 2026 Summary sheet). They sum to 77.451318, the residential sector weight to six decimals. They are residential sections expressed as shares of the whole index, not the material content of Saudi construction, and reading one as the other puts the material share out by eleven percentage points. The same sheet produced the standing line that Saudi build costs are a labour story: labour +2.75% against materials +1.21% in April 2026 are residential rates, where the whole-index pair is +2.81% and +1.16% (GASTAT, April 2026).

Omran arithmetic. GASTAT names its formula two ways in one document, the "Laspeyres equation" and the "Geometric Laspeyres Formula" (GASTAT). Its own numbers settle it. Weight the four July 2026 section indices (100.590500, 105.034299, 107.628165, 134.931510) geometrically by their Table 2 weights, exp(sum of w times ln I), and you get 103.9912920479984 against a published 103.9912920479984. Weight them arithmetically and you get 104.13, out by 0.14 index points. It reproduces again inside Energy, and again a level further down inside plant hire, where the five with-operator machine indices give 109.09113883238503 against a published 109.0911388323851. Any reader with a spreadsheet can check that in one line.

Where Saudi build-cost inflation actually sits

Plant hire, on a fifth of the weight. Equipment and machinery rental rose 4.53% in the twelve months to July 2026 against 1.80% for materials and 1.51% for labour, the fastest of the four sections in that print (GASTAT, Table 2) and the fastest in the April print too, at +5.12%. Labour's rate peaked in April at 2.81%.

Omran arithmetic. The exact decomposition of a geometric index is w times ln(I now / I twelve months earlier): 0.8663, 0.8427, 0.4496 and 0.0746 log points for materials, plant hire, labour and energy in July 2026. They sum to 2.2331 log points, and exp(0.022331) minus 1 gives +2.2581974%, the published headline to seven decimals. Scaled by (e^L minus 1)/L = 1.011245 so the parts sum to the published percentage: 0.8759pp, 0.8522pp, 0.4547pp, 0.0754pp. On the April print, multiplier 1.012178, plant hire gives 0.9615pp against labour's 0.8401pp of a published +2.4446%.

Materials edge July on contribution, 0.8759pp against 0.8522pp, and it takes 48.52% of the index to do it. Plant hire does almost as much on 19.01%. April is the cleaner test, because labour's rate was at its high for the year and plant hire still contributed more: 19.01% of the index moving 5.12% beats 29.93% of it moving 2.81%. A rate without its weight is not a story. Two of the ten labour lines GASTAT prices separately each month are cheaper than a year earlier in July, general labourer at minus 0.35% and builder at minus 0.47% (GASTAT, Table 2).

The operator premium: 43.55% of the index carries a wage

GASTAT prices the same five machines twice, hired with an operator and without, and publishes both rates. In the twelve months to July 2026 the operated rate rose faster in all five cases, by 1.53 to 8.04 percentage points. With-operator hire is 13.625783% of the whole index, bare hire 5.385156% (GASTAT, Table 2).

Ten published rows in Table 2 of the July 2026 workbook, and no one-month artefact: the April print repeats it five for five, with every gap wider.

Omran inference, not a GASTAT statement. The split is GASTAT's own design and the twenty rates are GASTAT's. The reading that the difference between them is the operator, and therefore a wage, is ours. On that reading, labour at 29.926% plus with-operator hire at 13.626% puts 43.55% of the index on a price containing a wage, against 48.52% for materials. Track Saudi pay exposure through the Labour section alone and you understate it by about a third. A hostile reader can name three other things the gap might hold: bundled fuel, contract length, a different rental segment. It holds five for five in two prints, and it remains an inference.

Escalating a benchmark when nobody publishes a SAR per sqm rate

No permitted source publishes a construction cost per square meter in Saudi Arabia, at any grade, in any city. GASTAT publishes an index with no price level behind it, REGA publishes property prices and rents rather than build costs, MOMRAH and NHC publish neither. So the deliverable is a method, and it is public.

Omran escalation method. Take a dated benchmark you already hold and multiply it by the ratio of the CCI general index in the target month to the index in the benchmark month, both read off one table. On the Table 2 and Summary vintage, a rate priced in July 2025 escalates to July 2026 by 103.991292 / 101.694822 = 1.022582, so +2.2582%. Table 3 of the same workbook carries a second vintage of July and returns 1.022595, which is why the clause must name its table and stay in it. Where the package is not shaped like the national basket, escalate each cost block on its own section index and re-weight to the project's cost plan. Call the output an Omran-method escalation, never a published rate. It feeds the build-cost line of a residual land value and the denominator of a yield on cost.

Three things break a clause drafted naively on it. The moves are steps, not trends: on Table 5, plant hire jumped 2.892% month on month in March 2026 and 0.956% in January, while the other five months of 2026 run between minus 0.062% and plus 0.333%; labour resets each July, 2.140% in 2024, 1.124% in 2025, 0.082% in 2026; energy moves in January and in no other month (GASTAT, Table 5). A twelve-month figure off this index is mostly the memory of three or four discrete resets.

The series is revised, which we found by diffing two releases. December 2025 was published at 101.795554 and now reads 101.853216 (GASTAT, December 2025 against the July 2026 file), so a contract indexed on that month with the number available in January 2026 escalated at +1.095% where the July file gives +1.152%. Both are GASTAT. The methodology sets a three-month maximum review window, yet the revised May 2025 value already sits in the February 2026 workbook released 17 March 2026, nine months after the reference month, on a change of 0.0002 index points. The window is not enforced even where nothing turns on it, so name your release.

And your package is not the national basket. Cement roof tiles carry 3.970541% of the whole index and ceramic wall brick 2.360035% on the July 2026 Table 2 weights (GASTAT), on a civils job that buys neither. GASTAT's metadata lists an "index for specialized construction activities" among its published variables and no release contains one, so nothing in the permitted universe prices Saudi infrastructure. In a market whose biggest contracts are civils, that is the most expensive way there is to misuse this number.

The costs this index never prices

Land is absent from the basket at every level, and the same office prices it on the same base year. GASTAT's Real Estate Price Index puts residential plots at 107.65 in Q2 2026, up 6.3% year on year (GASTAT REPI Q2 2026, released 19 July 2026). Compare sub-index with sub-index, never headline with headline: 70.6% of REPI by weight is bare land.

Omran arithmetic. Averaging the published monthly CCI values for April, May and June 2026 gives a Q2 2026 mean of 103.8537, which is 3.85% above the 2023 base and 2.58% above the Q2 2025 mean of 101.2416. That is the right side of the comparison when the other side is quarterly.

Over that quarter the land line ran 6.3% against 2.58% for the whole input basket. The one input this index does not measure moved about two and a half times as fast as everything it does (Riyadh land prices by district, Saudi real estate price index).

Set that against REPI's villa line, 94.89 in Q2 2026, down 9.7% year on year and 5.11% below the same base, and you have the sharpest margin statement public Saudi data allows: the finished villa transacts below its 2023 price while the inputs to build one cost 3.85% more. Margin squeezed from both ends in three years, on one office's own numbers. One caveat travels with that comparison: REPI's villa line is a transaction-price index across all vintages and locations, not a new-build price.

Three holes are left. Collection covers all 13 administrative regions and no regional index is published (GASTAT methodology), so Riyadh and Tabuk share one national number, and REGA's platform mirrors this series rather than producing a second (REGA). Design fees, supervision, finance cost and any series before 2023 are absent, so a benchmark priced in the 2020 to 2022 material shock needs a labelled assumption before it can be escalated. And this is not a tender price: the nearest public evidence of what contractors bid is Etimad's award record, 462,441 awarded competitions read on 29 August 2026, cumulative, with no stated start date and not openable without an account (Etimad).

That distance, between a published input index and the number a contract turns on, is where our work sits. Omran keeps the monthly series, the section and item detail, the revision history and the workings above in one dated table. Ask for the monthly table behind this page or scope an escalation review against your own benchmark.

FAQ

What is the Saudi construction cost index right now, and when is the next print out? 103.99 for July 2026 on a 2023 = 100 base, up 2.26% year on year, released 18 August 2026. Seven prints run from December 2025 to July 2026, each landing between the 15th and the 21st of the following month, and they are not consecutive: no January 2026 file could be retrieved. On that cadence the August print falls around 15 to 18 September 2026, which is our projection from the file headers and not a GASTAT commitment.

How much have Saudi construction costs actually risen since 2023? By 3.99% in total, roughly 1.3% a year, far below what market conversation assumes. On Table 4 of the July 2026 workbook the year-on-year rate sat inside a 0.65% to 1.15% band in every month of 2025, and the first half of 2024 was negative for six consecutive months, from minus 0.475% in January to minus 0.045% in June.

What is the construction cost per square meter in Saudi Arabia? No source in the permitted universe publishes one. Circulating SAR per sqm figures trace to private cost guides, to contractors' own marketing pages, or to content farms with no attribution at all. What is publishable is the escalation of a dated benchmark you already hold, with the arithmetic shown, as above.

Can I write this index into a contract escalation clause? Only with a named table and a named release. The July 2026 workbook disagrees with itself about July across two of its tables, December 2025 has already been revised from 101.795554 to 101.853216, and a May 2025 value moved nine months after the reference month against a stated three-month window. Never mix a level from one table with a rate from another.

Can I escalate a roads or earthworks package on it? No. It is a buildings index covering residential and non-residential building construction, and those two sectors exhaust its weight. GASTAT's metadata names an index for specialized construction activities that no release contains, and no permitted source publishes a Saudi infrastructure cost series at all.

Does REGA publish a different construction cost index? No. REGA's Real Estate Indicators Platform carries a Construction Cost Index panel badged "Beta Launch" and labelled "Data Source: General Authority for Statistics", running 2023-08 to 2026-07. Those strings sit in the HTML the platform serves. There is one Saudi construction cost index and REGA mirrors it.

Sources

• GASTAT, Construction Cost Index July 2026 workbook (index, sections, trades, Tables 2 to 5): link
• GASTAT, Construction Cost Index April 2026 workbook (April print, operator gaps, residential rates): link
• GASTAT, Construction Cost Index December 2025 workbook (revision evidence): link
• GASTAT, Construction Cost Index February 2026 workbook (dating the May 2025 revision): link
• GASTAT, Construction Cost Index June 2026 workbook (publication cadence): link
• GASTAT, Construction Cost Index page and FAQ (sector coverage, base year): link
• GASTAT, Methodology and Quality Report for the CCI (weights, formula, revision policy, regional coverage): link
• GASTAT, Real Estate Price Index Q2 2026 workbook (plot and villa sub-indices): link
• REGA, Real Estate Indicators Platform (GASTAT-sourced CCI panel, beta): link
• Etimad, public tender listing (462,441 awarded competitions, cumulative): link