00 · Why now: dated evidence and scenarios
An estimated 370,000 foreign office roles to assess for automation and productivity sharing.
This isn't a theoretical proposal about the future. Riyad Bank returned 150,000 hours to the business through automation. Mobily cut response times from 20 minutes to 6 seconds. Arab News now publishes in 50 languages via AI. AI is already working. The open question is who captures the gain.
estimated office-role pool derived from GASTAT/GLMM Q4 2025 data; automation feasibility is not measured
illustrative annual salary base: estimated roles × assumed SAR 3,900/month × 12; not observed firm payrolls
2025 fiscal breakeven estimate in the IMF August 2025 report. Its ~$62 oil assumption concerns 2026; these are dated projections.
Illustrative uptake scenarios for the estimated role pool; these are not measured adoption forecasts:
| Phase | Market % | Merges | Annual salary-value | Remittance-equivalent |
|---|---|---|---|---|
| Year 1 (conservative) | 1% | ~3,700 roles | ~SAR 173M | ~SAR 41M |
| Year 3 (assumed growth) | 5% | ~18,500 roles | ~SAR 864M | ~SAR 206M |
| At scale | 15% | ~55,000 roles | ~SAR 2.6B | ~SAR 617M |
Without a productivity-sharing mechanism
AI does not automatically raise wages. Firms can retain savings, reinvest them, lower prices, or share them with workers. This proposal asks how productivity gains could reach Saudi workers.
EstimateGASTAT Q4 2025 register, ISCO group 4: 60,227 + derived estimates for adjacent groups.EstimateWages per GASTAT non-Saudi services avg ~SAR 3,900/mo.Estimate2025 remittances reported by Argaam, citing SAMA; primary table not independently reproduced.EstimateIMF August 2025 projection: 2026 breakeven = $86.6/bbl; not a current observation.
01 · The gist
The idea in one sentence
This is a policy proposal, not an enacted program. When training and automation demonstrably release recurring costs, it proposes sharing 55% of the verified net saving with Saudi employees as salary increases. The calculator illustrates a gross cost pool; it does not deduct AI, transition, training, or additional employer costs. Direct state wage subsidy is assumed to be zero, which does not establish zero total public cost.
02 · Illustrative recurring cost
SAR 6,969/month: the calculator assumptions
A scenario using the calculator defaults. It is not a payroll-compliance calculation or a verified market-average cost.
| Component | Case 1: SAR 4,500 base | Case 3: SAR 13,000 base | Source |
|---|---|---|---|
| Base wage | 4,500 | 13,000 | Employment contract |
| Housing allowance (22%) | 990 | 2,860 | [Scenario] 22% of base |
| Labor fee | 800 | 800 | [Scenario] verify applicable fee |
| Health insurance | 400 | 400 | [Scenario] insurance allowance |
| EOS allowance (4.2% of base) | 189 | 546 | [Approximation] not a statutory settlement |
| GOSI cost proxy (2% of base) | 90 | 260 | [Proxy] actual contributory wage includes housing |
| Total | 6,969 | 17,866 | [Scenario] sum of these assumptions |
The calculator applies EOS and GOSI percentages to base pay only as simplified allowances. GOSI guidance includes cash housing in contributory wages. EOS entitlement depends on applicable wages, service and termination conditions. Reconcile actual payroll, fees and insurance before using any saving in a proposal.
An eligible industrial fee-exemption scenario: The official 17 December 2025 announcement cancels the expatriate levy for establishments holding an industrial license. The checkbox assumes eligibility and sets only that modeled fee to zero. At the default base salary the gross cost pool is SAR 6,169/month; the simplified VAT-minus-fee balance is about +SAR 4,275/year per role. This excludes other public costs and is not a fiscal-return forecast.
03 · The rule
The surplus, split three ways
So what do we do with this surplus?
A permanent raise to the employee’s salary, logged in Qiwa/Mudad. Not a bonus that can be clawed back.
Profit the firm keeps, so it benefits and repeats the move.
Finances the next one, so the program partly funds itself.
The worker takes the largest share. That was the point from the start: the benefit reaches the citizen's pocket, not just the firm's books.
New salary = old salary + 55% of the surplus. Why does the firm agree? Because it still earns a profit + a citizen employee who is harder to lose to a competitor.
04 · Not hypothetical: already happening
Three real cases from the Saudi market
The mechanism describes what's already happening, only without the surplus distribution. These cases show the missing piece.
What the mechanism describes: With automation support, the Saudi accountant now covers scope that used to require extra staff, and deserves a permanent raise from the surplus the firm freed. Without the mechanism: savings go to margin; employee stays at the same salary.
What the mechanism describes: The Saudi agent handles complex escalations while AI takes first contact and routine cases. Covering that larger volume deserves a raise from the savings. Without the mechanism: the firm freezes hiring or trims headcount; wages don't move.
What the mechanism describes: The Saudi content writer now acts as editor-in-chief and quality reviewer, covering the scope of translator, editor, and distributor. That deserves a permanent raise from those savings.
Cases sourced from public corporate announcements. Raise figures are computed from the mechanism formula (55% of freed surplus ÷ number of Saudis absorbing expanded scope). The named firms did not announce them.
05 · The budget in three scenarios
What does the budget look like at 50K, 100K, or 200K merges?
A narrow illustration of modeled VAT less the labor fee forgone. It excludes program administration, transition costs, existing spending and tax effects, and wider economic responses.
Calculator defaults, one retired role per merge: 70% non-industrial (about −SAR 4,770/year) + 30% eligible industrial-exempt (about +SAR 4,275/year). Rounded totals below use unrounded formulas:
| Scenario | Merges | Fee lost | Modeled VAT proxy | VAT less fee (70/30) | State-budget share |
|---|---|---|---|---|---|
| Conservative | 50,000 | −SAR 336M | +SAR 233M | −SAR 103M | Not estimated |
| Base | 100,000 | −SAR 672M | +SAR 466M | −SAR 206M | Not estimated |
| Ambitious | 200,000 | −SAR 1,344M | +SAR 933M | −SAR 411M | Not estimated |
vs. HRDF
SAR 0
Direct state wage subsidy assumed in this proposal. Administration, training support, enforcement and transition costs are not modeled, so this does not establish a cost advantage over HRDF programs.
Wider economic effect
Not modeled
No calibrated multiplier is established here. Imports, overseas AI costs, worker displacement and changes in consumption must be assessed before forecasting a national benefit.
ScenarioFigures are calibratable design assumptions.ScenarioAssumed labor fee SAR 800/mo × 12 = SAR 9,600/yr; applicable fees and exemptions vary.ScenarioVAT proxy: 15% × worker allocation × 12 × 70% assumed local taxable spend, treated as pre-VAT. Not measured incremental revenue.FactOfficial announcement, 17 December 2025: levy cancelled for industrial-license establishments; confirm eligibility.
06 · Remittances: a dated baseline and scenario
What does the 25% assumption mean?
Argaam reports SAR 165.51B in expatriate remittances for 2025, citing SAMA. The primary annual table was not independently reproduced in this review. A 25% share is a design assumption, not an observed automatable share.
The following tier ranges are illustrative assumptions, not an official decomposition of the 2025 national total. A survey of Indian expatriates cannot establish shares for every nationality or occupation:
For scale illustration only, 25% of the reported 2025 total is about SAR 41B/year. This does not establish how many roles can be automated, how much payroll can be saved, or how much money remains in Saudi Arabia.
The proposal focuses on selected office workflows and excludes construction, manual, care and domestic roles from its intended scope. These exclusions do not demonstrate that 75% of national remittances is unaffected.
Economic interpretation: A retired role’s assumed remittances are an outflow equivalent, not money automatically retained locally. Saudi households can save or spend abroad; AI vendors and imports can create new outflows. The calculator’s 70% local-spend share is a fixed scenario assumption, not measured behavior or proof of a stronger multiplier.
EstimateDated secondary report: Argaam, 2025 total SAR 165.51B, citing SAMA; checked 2 October 2026.ScenarioTier ranges are illustrative; PMC8464826 and World Bank/KNOMAD do not verify this national occupational distribution.Estimate~199,309 formal clerical workers, per the GASTAT/GLMM private-sector administrative register.
07 · Safeguards backed by data, not words
Sham Saudization exists. Here's what actually detects it
We say it before any critic does: gaming is possible. But it's documented, costly, and detectable with tools that exist today.
MHRSD Saudi Arabia, Q1 2026: invalid employment violations confirmed from 91,000 suspected cases screened by smart monitoring + field visits = 14.8% of the suspect sample. 7,200+ visas cancelled/withdrawn.
Singapore, FCF 2016–2020: employers scrutinized · 3,200 EP applications rejected/suspended/withdrawn · 4,800+ Singaporeans hired post-investigation. Detection: data analytics + single-nationality concentration as a risk flag.
UAE, MOHRE H1 2025: sham Emiratization cases detected via field inspection + digital monitoring system. Proves Gulf-scale detection is operationally feasible.
What detection tools already exist in Saudi Arabia and actually work:
Employment contract + salary + GOSI registration are necessary, but not sufficient. Sham Saudization is defined as registered in GOSI and receiving salary with no evidence of actual work, so GOSI alone does not prove production.
Inspector reviews: employment contracts + staff IDs + attendance records + payrolls + worker statements + contact with registered-but-absent employees. Fine: SAR 25,000 per registered person, and it accumulates.
The decisive layer: actual production certificate. The mechanism ties the premium to a documented outcome on real work. Not attendance, not registration.
Production-proof specifications per case type:
Transaction volume processed + exception rate from system logs (e.g., Riyad Bank → 240s to 45s per document, verified from system log not manual report).
Case handling rate + quality scores + escalation volume from CRM (e.g., Mobily → case-processing time extracted directly from Sprinklr).
Content pieces produced + AI-assist audit trail (AI vs. human-edit ratio) from content tool (e.g., Arab News → publication rate supported by CAMB.AI).
08 · Preventing entrenchment
Keeping the AI premium Saudi
- 1
The premium is for the Saudi only: the surplus (the 55%) is released only when a Saudi takes the expanded scope. A foreigner covering extra work = zero premium. It kills the firm's financial incentive to make the foreigner the hub of multi-role work. (This is the core; the rest reinforces it.)
- 2
One permit = one occupation: tie the foreign permit to a single occupational classification (the 2025 skill-based permit system already classifies). Want them on a second occupation? Not allowed on the same permit: either hire a Saudi or buy a second, costly permit. It limits role-stacking structurally, not by banning skill.
- 3
Nitaqat credit only if the consolidator is Saudi: a merged role earns a Nitaqat point only when a citizen covers it. A foreigner consolidating = zero Nitaqat benefit.
- 4
Tiered fees on foreign-worker density: the higher the foreign density and wage mass in a cost center, the higher the fees. A tax on entrenchment, not a ban on skill. It makes the expanded foreigner expensive, and firms respond to cost faster than countries respond to bans.
- 5
A shadow-Saudi on every sensitive foreign role: senior AI experts stay welcome, but each one above a threshold needs a shadow-Saudi + objective knowledge-transfer milestones (certifications the shadow must pass) tied to visa renewal. Expertise transfers instead of entrenching.
- 6
A government procurement preference: government and PIF contracts favor merged roles staffed by Saudis, especially in sensitive work. The same logic as well-known local-content rules.
- 7
The 'certified consolidator' status is a citizen credential: anyone can learn AI, but the certificate that releases the premium and Nitaqat credit (SAMAI/Tuwaiq) is a citizen track. You distribute the benefit-bearing certificate; you don't police learning.
- 8
Merged sensitive roles require Saudi access: where AI lets one person hold roles touching sensitive data or control systems, require a citizen or Saudi-controlled access, justified by data protection (PDPL) and critical infrastructure rather than by nationality.
- 9
Mandatory reporting by nationality: large firms report roles-per-head and output-per-head split Saudi/non-Saudi. What you don't measure you can't manage. The data stays regulatory and is never used for naming-and-shaming.
- 10
A foreigner's multi-role tenure is temporary + time-capped: if a foreigner holds a merged super-role as a stopgap, a 24–36 month cap during which a ready, certified Saudi takes over, and the permit is non-renewable afterward. Entrenchment turns from permanent to temporary by design.
08a · Legal classification: the regulatory anchor for Lever 2
MHRSD Work Permit Skill Classification, May 2025
Lever 2 in §08 says 'tie the foreign permit to a single occupational classification.' That classification exists in law: Ministerial Decision 4602.
| MHRSD Tier | Arabic name | Definition (ISCO major groups) | Role in mechanism |
|---|---|---|---|
| Top | High-skilled | Groups 1–3: managers, professionals, technicians, with a salary floor and professional certification | Exempt (frontier tech elite) |
| Skilled | Skilled | Groups 4–8: clerical support, services/sales, craft, operators, with a salary floor and certification | In scope (especially 4xxx codes) |
| Basic | Basic | Group 9: elementary occupations | Out of scope (manual elementary work) |
The mechanism's core target is ISCO group 4xxx, clerical support workers. These codes define the jobs by legal name:
| SSCO Code | Occupational unit | In scope? |
|---|---|---|
| 4110 | General administrative clerks | Yes |
| 4120 | General secretaries | Yes |
| 4131–4132 | Typists & data entry operators | Yes |
| 4211 | Bank tellers & related clerks | Yes |
| 4222–4225 | Contact-centre & inquiry clerks | Yes |
| 4311–4313 | Accounting, payroll & insurance clerks | Yes |
| 4321–4323 | Inventory, production & transport clerks | Yes if back-office operations |
Legal anchor for Lever 2 §08
Lever 2 is anchored to Ministerial Decision 4602, Work Permit Classification by Skill Categories (May 28, 2025): a foreign permit is tied to the occupational major group classified at issuance. Lever 2 doesn't create a new restriction. It applies an existing one and links it to the condition that a Saudi absorbs the expanded scope. The elite carve-out uses MHRSD's own 'High-skilled' category rather than an internal term, which makes it directly embeddable in regulation.
09 · Calculator
Run your own numbers
Adjust the assumptions to explore a gross cost pool, the proposed 55/35/10 allocation, an illustrative salary increase, a remittance equivalent, and a simplified VAT-minus-fee balance.
Inputs
Cost assumptions (editable)
Results
Illustrative scenario, not a forecast or statutory payroll calculation. Costs and remittance share are editable; 70% local taxable spend and 15% VAT are fixed assumptions, treating spend as pre-VAT. Existing spending and tax effects are omitted. Gross cost is allocated without deducting AI, transition, training or added employer contributions. Remittance equivalents are not proven local retention, and VAT less fees is not the full fiscal impact.
10 · Objections
Meeting the hardest objections head-on
Anyone reading this critically will fire these five. Here are the answers, stated honestly, including where the hole stays open.
01Firms will fake the AI certificate: a two-day course and done.
The premium is paid only against a certificate backed by real production output on real work, not attendance. The approval that releases the money is tied to a documented outcome: higher salary plus expanded scope logged in Qiwa/Mudad. So faking it means faking sustained real output. That is far harder, and it surfaces on audit.
02Rent-a-Saudi front-man + a foreigner working behind the curtain, and the whole thing collapses.
This is the most dangerous hole, and we state it plainly rather than deny it, because it breaks more than one safeguard at once. It's shrunk by department-level foreign-headcount monitoring, output-per-head split Saudi/non-Saudi, and GOSI plus real-attendance linkage. But it never goes fully to zero. Anyone who tells you they can eliminate it is lying. The realistic goal is to make it costly and hard, not impossible.
03Friction on senior foreign AI talent fights the goal of attracting frontier talent.
True, which is why the senior expert stays welcome, with no Saudization premium attached. But they’re paired with a shadow-Saudi who absorbs the knowledge, objective transfer milestones (certifications the shadow must pass) tied to visa renewal, and a 24–36 month cap. The aim is to transfer expertise, not expel it. The carve-out for frontier roles is explicit and written down.
04This is protectionism in new clothes, and it'll land you in trade-law trouble.
No. We bar no foreigner from work or from learning AI. We distribute a productivity premium and a certificate to citizens (something every country does), and we plug remittance leakage. The whole frame is anchored to economic capture, permit scope, and security. It is never anchored to "nationality + skill." The procurement-preference and sensitive-access levers sit under legal review (GATS national treatment, BITs, ILO equal-pay) before any external circulation.
05The net budget is negative, so why do this at all?
We concede it: ~SAR 4,750/year per merge at steady state. But it isn't sold on the budget; it's sold on growth, Saudization, and household wealth. For industrial-license firms (fee-exempt) the net is already positive. And any future personal-income tax flips the whole equation in the state's favor.
Objection: 'This is just dressed-up protectionism'
We didn't invent the idea. Here's who does something closest
The answer isn't 'every country protects its workers.' The answer is more precise: every advanced economy makes productivity gains from technology a right for local workers, in different forms.
| Country | Policy name | Mechanism | Documented outcomes | Similarity to Premium |
|---|---|---|---|---|
| France | Mandatory profit-sharing (participation) | Firms with 50+ workers legally required to distribute a share of net profit to employees | +1.8 ppt in labor share · −1.4 ppt in profit share · no negative investment effect (NBER) | Closest payment analogue. Difference: distributes general profit, not AI savings specifically |
| Germany | Works councils + Qualifizierungsgeld | Works councils involved in any fundamental change to work methods or manufacturing. Qualifizierungsgeld: 60–67% wage replacement during training, employer-funded | Mature transition architecture: compensation, retraining, enforceable social dialogue | Strongest tech-displacement analogue. Difference: negotiated, not a statutory savings-split |
| Singapore | Fair Consideration Framework + COMPASS | Firms must advertise locally and genuinely consider local candidates before any foreign EP is approved. COMPASS scores EP applications partly on diversity and local PMET support | 1,200+ employers scrutinized · 3,200 EP applications rejected · 4,800+ citizens hired post-investigation (2016–2020) | Closest anti-entrenchment analogue. Difference: a process mandate, not an outcome; the Premium transmits wages directly, which is stronger |
What is genuinely novel in the Saudi proposal: No other country mandates a fixed share of AI-automation savings specifically as a permanent salary raise for local workers. The novel design is the combination of three principles in a single mechanism with a defined split ratio: French mandatory sharing, German transition architecture, and Singapore anti-entrenchment targeting.
11 · Who wins, who pays
The stakeholder ledger
No assembling the answer across eight sections. Every party's net sits in one row.
| Party | What they get / pay | Net |
|---|---|---|
| Saudi worker | +55% of the surplus as a permanent raise · an AI certificate · broader job scope | Large + |
| The firm | +35% retained margin · a stickier citizen workforce · a procurement edge | Moderate + |
| The state | −lost labor fee · +VAT on citizen spend · +Saudization & household wealth · pays SAR 0 toward the raise | Depends on assumptions; full impact unmodeled |
| The expat worker | The office role is phased out gradually · senior experts retained via carve-out + time cap | − for automatable roles |
| The economy | Potential productivity and skills gains; national spending, remittance and employment effects require evidence | Unmodeled |
Manual, care and domestic roles are outside the proposal’s intended scope. Indirect employment and demand effects are not modeled.
Stakeholder ledger: gender and regional distribution note
Saudi women fill 55.7% of insured clerical roles, yet they start from lower pay
Saudi female LFPR, Q4 2025 (above Vision 2030's 30% target)
of GOSI-registered clerical support workers are Saudi women, the most mechanism-eligible occupation group
of Saudi GOSI subscribers in 3 regions: Riyadh, Eastern Province, Makkah. Without a regional design layer, the benefit concentrates in major metros.
Design alert: The mechanism’s fixed-percentage formula (+29–35%) preserves the relative gender wage gap and doesn’t close it. Saudi male average: SAR 11,836. Saudi female average: SAR 8,939 (24.5% gap). At +35%: men gain +SAR 4,143, women +SAR 3,129. Same percentage, wider absolute riyal gap. If closing the gap is a priority, the mechanism needs a minimum cash-uplift floor, not just a percentage split. [Fact] GASTAT Q4 2025 wage tables · IMF Art.IV 2025 report.
12 · Sources and verification limits
Every figure, where it came from, and its label
Reviewed 2 October 2026. Dates belong to the cited period or projection. The remittance baseline is a secondary report; unlinked historical claims below have not been freshly verified. Scenario assumptions do not establish feasibility, legal adoption or economic outcomes.
Argaam reports expatriate remittances of SAR 165.51B in 2025, about 15% above 2024, citing SAMA. This review did not independently reproduce the primary annual table.
Argaam secondary report on 2025 data · checked 2 October 2026 · Argaam
Labor Law Art. 84: EOS = ½ month/yr for first 5 years, 1 month/yr thereafter. Monthly accrual at SAR 4,500 (years 1–5) = SAR 187.5 ≈ 190.
MHRSD · Saudi Labor Law Art. 84
GOSI: Occupational Hazards Branch = 2% of wage, employer-only, applies to all nationalities. Annuities/pension applies to Saudis only.
GOSI employer FAQ; contributory earnings include basic wage and cash housing GOSI employer FAQ
Official 17 December 2025 announcement: the expatriate levy is cancelled for establishments licensed under an industrial license. The calculator assumes eligibility; it does not assess it.
Umm Al-Qura / Saudi Press Agency, 17 December 2025 Official announcement
MHRSD Q1 2026: 91,000 suspected sham-Saudization cases examined · 13,509 invalid-employment violations confirmed = 14.8% of suspect sample · 7,200+ visas cancelled/withdrawn.
MHRSD Q1 2026 report
Riyad Bank + SS&C Blue Prism August 2023: 200 processes automated, 150,000+ hours returned in 6 months, document processing 240s → 45s (−81%).
PR Newswire · official release August 2023
Mobily + Sprinklr/Cognigy.AI: first-response 20 min → 6 sec (−99.6%) · case-processing −68% (3m55s → 1m16s) · covers X, WhatsApp, Apple Business Chat, Instagram.
Sprinklr case study · Cognigy.AI customer story
Arab News + CAMB.AI October 2025: content now available in 50 languages, reaching 6.5B people (80% of the world's population).
Arab News · official announcement October 2025
Ministerial Decision 4602, Work permit classification: three tiers (high-skilled, skilled, basic) based on Saudi Standard Occupational Classification, salary floor, and professional certification. May 28, 2025.
MHRSD · hrsd.gov.sa
Dated IMF August 2025 estimates/projections: Saudi fiscal breakeven ~$92.3/bbl in 2025 and ~$86.6/bbl in 2026; oil-price assumption ~$62.4/bbl for 2026. Not current realized values.
IMF Country Report No. 25/223, published 4 August 2025 IMF 2025 report
The '50% or 52% AI reskilling efficacy rate' figure is confirmed as a distortion. The original figure was: 52% of SAMAI participants are women (demographic split, not efficacy rate). Correct HRDF figure: ICT employment sustainability rose from 49% to 81% (2020–2025). WEF 2023: 52% of Saudi employers prioritized AI/big data for reskilling.
SDAIA/MoE SAMAI announcement · HRDF 65,000 ICT jobs announcement · WEF Future of Jobs 2023
Illustrative tier propensities: construction 75–80%, domestic 50–65%, clerical/service 30–45%, senior tech 10–25%. These sources do not verify this decomposition of national remittances or an automatable 25% share.
Background only: PMC8464826 · World Bank/KNOMAD · GASTAT HICES 2023
Singapore FCF 2016–2020: 1,200+ employers scrutinized · 3,200 EP applications rejected/suspended/withdrawn · 4,800+ Singaporeans hired post-investigation. Pre-selection penalty: barred from foreign hiring up to 24 months.
Singapore Ministry of Manpower · FCF enforcement statistics
France, mandatory profit-sharing: +1.8 ppt in labor share, −1.4 ppt in profit share. No documented negative effect on investment or productivity.
NBER Working Paper on the French 50-employee threshold
GASTAT Q4 2025: Saudi female LFPR 34.5% · female unemployment 10.3% · 55.7% of GOSI clerical-support subscribers are women · Saudi male avg wage SAR 11,836 / female SAR 8,939.
GASTAT Q4 2025 labor-market register · IMF Art.IV 2025 report