Key takeaways
- Only about 1 in 4 companies (26%) see their AI returns accelerating year over year. Google Cloud calls them "AI ROI leaders".1
- Leaders do three things more than others: they give AI a clear owner (48% vs 27%), make AI training part of every role (38% vs 18%), and build AI into core business processes (48% vs 27%).1
- Security and compliance is the #1 enabler for scaling AI (46%). In Saudi Arabia, that starts with the Personal Data Protection Law (PDPL).1,2
- Only 12% of companies measure AI with a formal value-tracking framework. Measuring is the cheapest advantage available.1
- To measure AI ROI: record a baseline for each process, build, then report hours, cost and return every month.
What is the ROI of AI for Saudi companies?
There is no single number. AI pays back where it is built into work that happens every day, such as approvals, invoices, branch reports and customer requests, and where one person owns the result. It rarely pays back as a tool that each team uses its own way.
The direction is clear, though. In Google Cloud's 2026 survey of 2,403 executives, 86% agree that AI has let them grow revenue or output without operating costs rising at the same rate, and 97% plan to spend more on AI next year.1 The question has moved from "does AI work?" to "where does it pay, and how much?". Google searches for "AI ROI" rose about 1,400% in twelve months.1
Who gets the most return from AI
Asked how their financial returns from AI changed over the last twelve months, executives split into four groups.1
Out of every 100 companies
How financial returns from AI changed over the last 12 months
- 26%Accelerating: the AI ROI leaders
- 58%Steadily increasing
- 16%Inconsistent, flat or too early to tell
- <1%Declining
Show the numbers
| Returns from AI | Share |
|---|---|
| Accelerating (AI ROI leaders) | 26% |
| Steadily increasing | 58% |
| Inconsistent, flat or too early to tell | 16% |
| Declining | <1% |
Source: Google Cloud, The ROI of AI 2026 (2,403 executives).
Almost nobody is losing. The gap is between companies whose returns compound and companies whose returns stall after the first wins.
Three habits of AI ROI leaders, in a Saudi company
Compared with everyone else, the companies whose AI returns are accelerating stand out on three habits.1
AI ROI leaders vs everyone else
Share of companies with each habit
- AI ROI leaders
- Everyone else
Clear owner for AI decisions
AI training built into every role
AI built into core processes
Show the numbers
| Habit | Leaders | Others |
|---|---|---|
| Clear owner for AI decisions | 48% | 27% |
| AI training built into every role | 38% | 18% |
| AI built into core processes | 48% | 27% |
Source: Google Cloud, The ROI of AI 2026 (2,403 executives).
1. They give AI a clear owner
48% of AI ROI leaders say ownership and decision rights for AI agent work are "extremely clear", against 27% of other companies.1 In many Saudi groups, AI sits between IT, the CEO's office and each department, so decisions stall. Name one owner with a budget and the right to say no, and write one AI direction for the whole company.
2. They make AI training part of every role
38% of leaders have ongoing, required AI training built into roles, against 18% of other companies.1 A one-off workshop is not enough. Train each team on its own real work, in Arabic and English, and make AI use part of the job rather than an optional course.
3. They build AI into core processes
48% of leaders have embedded AI into core processes and revenue, or use it for new business models, against 27% of other companies.1 Start where volume is high and the rules are clear: invoices, approvals, branch reports, customer requests. Build on the systems the company already runs instead of adding another tool.
Where the return shows up
Faster decisions have overtaken productivity as the most-cited result of AI investment.1
Where AI is measurably paying off
Share of executives reporting each outcome
Show the numbers
| Executives | |
|---|---|
| Faster strategic decisions | 55% |
| More workforce capacity | 52% |
| Faster innovation cycles | 47% |
| Higher customer lifetime value | 46% |
| Revenue from new products or services | 43% |
| Risk intelligence and prediction | 43% |
| Revenue growth in existing offerings | 40% |
| Competitive differentiation | 36% |
| Margin expansion | 31% |
Source: Google Cloud, The ROI of AI 2026 (2,403 executives).
Agents are the main driver: 94% of respondents say AI agents contribute to both cost savings and revenue.1 For a Saudi leadership team, that means live numbers instead of month-end reports, and agents doing the repetitive steps inside existing systems.
What it takes to scale AI beyond a pilot
Asked what most helped them move AI from pilot to measurable value, executives put security and compliance first.1
What helped companies scale AI
Share of executives naming each capability
Show the numbers
| Executives | |
|---|---|
| Security, compliance and regulatory readiness | 46% |
| Redesigning workflows to put AI into core processes | 43% |
| Workforce training and change management | 41% |
| Cloud or infrastructure modernization | 41% |
| A central AI operating model | 40% |
| Clear executive sponsorship and accountability | 39% |
| Fewer tools and vendors | 39% |
| Unified or modernized data infrastructure | 37% |
| A dedicated ROI or value-tracking framework | 37% |
Source: Google Cloud, The ROI of AI 2026 (2,403 executives).
In Saudi Arabia, the first item starts with the Personal Data Protection Law (PDPL), overseen by SDAIA. It sets rules for processing personal data, including sending it outside the Kingdom.2 Staff pasting customer data into foreign AI tools is the fastest way to create risk. Clear rules on which tools are approved, which data may go into them and who approves sensitive actions are what let AI scale safely.
How to measure AI ROI
Most companies judge AI by operational improvements or direct financial results. Only 12% use a formal value-tracking framework or executive reporting.1
How companies measure AI today
Main way success with AI is measured
Show the numbers
| Companies | |
|---|---|
| Operational performance improvements | 36% |
| Direct financial outcomes | 21% |
| Strategic business KPIs | 18% |
| Time to value or speed of return | 13% |
| A formal value-tracking framework or executive reporting | 12% |
Source: Google Cloud, The ROI of AI 2026 (2,403 executives).
A simple method is enough:
- 1Record a baselineHours, cost, errors and cycle time for each process, before building.
- 2Price the buildAgainst the baseline. If it cannot pay back, do not build it.
- 3Launch and measureThe same numbers, every month.
- 4Report the returnHours saved, cost and return per department, to leadership.
Steps 3 and 4 repeat every month.
AI ROI = (value gained − total AI cost) ÷ total AI cost value gained = hours saved × loaded hourly cost + costs removed + extra revenue total AI cost = build + licences and model usage + training + running
Count model usage (tokens) as a running cost. Newer models often do the same work for less, so moving to better, cheaper models as they ship is part of keeping the return up.
Illustrative example
A finance team spends 600 hours a month entering invoices, at a loaded cost of SAR 80 an hour. Automation removes 400 of those hours: SAR 32,000 a month, or SAR 384,000 a year. If the build, licences, training and running cost come to SAR 120,000 in the first year, the first-year ROI is (384,000 − 120,000) ÷ 120,000 = 2.2, a 220% return. The numbers are made up to show the method; the real ones come from the baseline.
First-year return, illustrative
SAR thousands. Made-up numbers to show the method.
ROI = 264 ÷ 120 = 2.2, a 220% return
From our field work in Saudi Arabia
In our audit of a Saudi multi-branch operator, we read real work files across the business before proposing anything3:
- 8departments reviewed
- 135manual processes found
- 51workflows scoped for phase one
- ~7xestimated return on the program price
The 7x is an estimate from the audit, priced against the baseline. Monthly reports after each build confirm or correct it.
Where to start: 30 days for a Saudi leadership team
- Week 1Own itName one AI owner. Agree one AI direction for the company, on one page.
- Week 2Measure itList the ten highest-volume manual processes. Record a baseline for three of them.
- Week 3Govern itSet data rules: which AI tools are approved, what data may go into them, who approves sensitive actions.
- Week 4Train and scopeTrain each team on its own work, and scope the first build with a payback target.
Questions about AI ROI
What is a good ROI for AI?
There is no fixed benchmark, because it depends on the process. Set a payback target for each process before building and only build what meets it. In our audit of a Saudi multi-branch operator, the estimated return was about 7x the program price.
How do you calculate AI ROI?
AI ROI = (value gained − total AI cost) ÷ total AI cost. Value gained is hours saved times loaded hourly cost, plus costs removed and extra revenue. Total AI cost is build, licences and model usage, training and running costs.
Why do most AI projects fail to show ROI?
Usually there is no clear owner, every team uses different tools, AI sits outside core work, and no baseline was recorded, so nobody can prove the gain. The companies with the fastest-growing returns fix ownership, training and core processes first.
What are AI ROI leaders?
In Google Cloud's 2026 survey of 2,403 executives, AI ROI leaders are the 26% of companies whose financial returns from AI are accelerating year over year.
Does PDPL stop Saudi companies from using AI?
No. The Personal Data Protection Law sets rules for personal data, including transfers outside Saudi Arabia. Companies can use AI by keeping personal data in approved systems, controlling what goes into outside AI tools, and having a named person approve sensitive actions.
Where should a Saudi company start with AI?
Name one AI owner, agree one direction, record baselines for three high-volume processes, set data rules, and train teams on their own work. Then build the first workflow with a payback target.