Saudi Arabia's Year of AI Will Be Won on the Frontline

By Waseem Afzal, Founder, FAST

Saudi Arabia has committed $9.1 billion to artificial intelligence. HUMAIN is building capacity across data centres, cloud infrastructure, models and applications, and the Cabinet has named 2026 the Year of Artificial Intelligence. Yet the businesses that will ultimately decide whether that investment pays off have, for the most part, fewer than fifty employees, no data science team, and no AI strategy at all.

That is not a weakness in the plan. It is the plan's real test. Because a Year of AI is not won by the biggest balance sheet. It is won by whoever closes the gap between national ambition and daily operational reality fastest. For Saudi Arabia's small and medium businesses, that gap is where they either get left behind or leapfrog.

The customer got there first

Something unusual has happened in the Kingdom, and it deserves more attention than it has received. Technology adoption normally flows from enterprise to consumer: companies invest, and customers catch up. In Saudi Arabia, AI has run the other way.

The Communications, Space and Technology Commission found that 45.2% of internet users were using AI tools in 2025, more than double the 21.5% recorded a year earlier. Of those users, 80.8% use AI to search for information. Deloitte's consumer survey points in the same direction, with adoption at 66%, up from 49% the previous year. Saudi consumers, in other words, adopted AI faster than most Saudi businesses did.

We can already see this shift arriving on brands' doorsteps. Across our client work in the Saudi market, we now see between 12% and 15% of traffic influenced by large language models, and it is growing far faster than traditional organic channels. A customer who researched a purchase through an AI assistant and got a clear answer in seconds will not wait patiently in a WhatsApp queue or tolerate a three-day email turnaround. Preparing for AI-led discovery is no longer a luxury for a brand. It is where the customer already lives.

A decade ago, we told ourselves that all marketing would become digital. It did. The next decade's version is already visible: all digital will be AI-enabled.

Two targets, one project

There is a second reason the SMB layer matters more than the coverage suggests, and it is written into the national strategy itself. Vision 2030 aims to raise the contribution of small and medium enterprises to GDP from around 20% to 35%. Progress has been real but gradual, and the remaining distance cannot be closed by headcount growth alone in a tight labour market. It requires a step change in productivity per enterprise, across more than a million businesses.

AI is the only instrument available at that scale. Which means the Year of AI and the 35% SME target are not two separate workstreams of the national vision. They are the same project. The data centres and sovereign models are the supply side. Adoption inside over a million small businesses is the demand side. Call the distance between them the absorption gap: the space between installed capacity and deployed capability. National AI strategies rarely fail on supply. They fail on absorption.

What it looks like from inside

I have lived that gap from the inside. I run a bootstrapped, marketing-first adtech company headquartered in Dubai, with a significant share of our work serving brands across the Saudi market. When we started building AI into our operating system 18 months ago, it began with a very ordinary growth problem: as the client base grows, the work around the client grows too. More reporting, more checks, more follow-ups, more administration. The traditional answer is to add more people to carry the load. We wanted a better way to grow.

Today, our internal operating system includes more than 180 AI agents across six platforms. Each was built around a real problem someone in our business faced every day. One example gives a sense of what that means in practice. We built an agent system we call AI CAST. It monitors trending topics, builds an AI avatar, writes the script, runs the full end-to-end content production, and publishes to social platforms. That process used to take a team days. It now happens in minutes. The people who once carried that production load now spend their time on judgement: which topics matter for which client, which angle earns attention, what the output should never say.

We made our share of mistakes along the way. Early on, some agents had been running for around a year before we properly owned the question of what this change meant for the people working alongside them. That was a miss. People naturally have questions when the way they work starts changing, and leaders need to create the space to answer them honestly.

Here is the fact I find most useful in those conversations: we employ more people today than when we started. We did not hire fewer people because of AI. We stopped hiring people into work nobody wanted to do. AI absorbed the administrative weight that scaling used to force onto every new hire, and hiring continued anyway, into roles built on the things machines do not supply: client trust, relationships, better sales conversations, judgement.

The biggest lesson has been that the technology was rarely the hardest part. The harder work was understanding our own operation well enough to know where AI could help, where people needed to keep their judgement, and how the two could work together.

The small business advantage

That lesson travels, and it is good news for the smallest firms in the Kingdom. Most businesses can now access broadly the same AI models. A trading company in Riyadh reaches roughly the same intelligence as the largest enterprise in the country. What cannot be bought is what each business already owns: the knowledge of why a particular supplier fails in Ramadan, which customer question signals a sale, what holds up a payment, where the same mistake keeps showing up.

For the first time, a small business's structural disadvantage, capital, has been decoupled from its structural advantage, proximity to its own knowledge. The model produces a competent first answer for everyone. The company that feeds it real operational context, and keeps a person accountable for the outcome, gets a better one. That is the mechanism by which SMBs do not merely keep up in the Year of AI, but leapfrog.

The starting point is deliberately unglamorous. Pick one recurring job: sorting inbound enquiries, checking a delivery issue, drafting a tender response, understanding the day's sales. Use the tool. Keep a person accountable for the result. Learn what works. Then build from there. Progress over perfection.

Saudi Arabia's national investment makes this easier than it has ever been. Local capacity and locally relevant models bring access closer to businesses of every size. The next layer is practical: tools businesses can trust, help improving one workflow at a time, and clear rules for using customer and business data responsibly.

The measure that matters

The Year of AI deserves to be measured by more than investment totals and computing capacity. Those figures show national intent, and the direction is set. But in 2027, the better question will not be how many data centres were built. It will be how many invoices were chased, tenders drafted, deliveries traced and customers answered by businesses that could never afford to do it before, and whether their customers felt the difference.

The infrastructure is being built. The ambition is real. What remains is absorption: turning national capacity into everyday capability, one business and one workflow at a time.