Saudi Arabia's artificial intelligence programme has been read as a data centre and language model story. The more consequential deployments are physical: PIF-owned Alat manufacturing industrial robots in Riyadh, HUMAIN building digital twins for manufacturing and logistics, and finishing and reinforcement robots arriving on Saudi construction sites.
The Saudi artificial intelligence programme is usually described in terms of compute and models. The deployments most likely to change an industrial profit-and-loss account over the next five years involve machines with actuators.
Alat, the Public Investment Fund manufacturing company established in February 2024 and chaired by Crown Prince Mohammed bin Salman, partnered with SoftBank Group to invest up to $150 million in an industrial robot manufacturing hub in Riyadh, and the venture moved to exporting Saudi-made industrial robots during 2025. The technology being licensed into it is self-programming automation intended to cut changeover and downtime rather than to generate text. HUMAIN, for its part, has agreed to deploy NVIDIA's Omniverse platform as a multi-tenant system in the Kingdom to support simulation, optimisation and operation of physical environments, aimed at manufacturing, logistics and energy customers building digital twins of real plant. HUMAIN has separately agreed a long-term collaboration with Applied Intuition on autonomy in Saudi Arabia, beginning with driverless trucking and with stated ambitions across ports, mining and other industrial settings.
What all of that has in common is that the value is realised by a physical asset behaving differently, not by a screen producing an answer. And that is precisely the category of application arriving on the Big 5 Construct Saudi floor this week.
The construction robots being shown in Riyadh are not general-purpose machines. They are single-task devices aimed at work that is repetitive, unpleasant or hard to staff. Shenzhen-based Legend Robot builds putty and latex paint spraying robots with 3.3-metre and 6.2-metre working heights, controlled from a tablet interface that requires no programming, and says its machines have covered more than 10 million square metres of construction to date. DaFang AI, also based in Shenzhen with offices in Singapore and Hong Kong, builds autonomous building surface finishing robots that combine machine vision with low-speed self-driving, and reports 128 patent filings with 81 granted. Neither company is selling intelligence in the abstract. Both are selling square metres per shift.
The same logic runs through reinforcement processing. Automated cut-and-bend lines, mesh welding plants and cage welders from suppliers including Schnell and Progress Maschinen & Automation now take bar schedules directly from building information models, and Progress has been adding machine-assisted quality control to its mesh welding lines. That is applied vision and control software doing a job that would otherwise consume a large steel-fixing crew — in a market that consumed around 15.5 million tonnes of rebar in 2025.
The economic case for physical AI in Saudi Arabia is not primarily a cost case. It is a capacity case. The Kingdom's construction workforce exceeded 3.4 million at the end of 2025 and is forecast to keep growing toward the end of the decade as giga-project delivery, the 2034 World Cup stadium programme and the Riyadh infrastructure build reach their peaks together. Against that, the sector is carrying a shortage of skilled workers running into the hundreds of thousands, with specialist roles staying open for three to six months. Automation is being bought to fill a gap that hiring cannot close on the required timeline.
The factory side of the argument is further advanced than the site side. Saudi manufacturers have been installing industrial robots and automated cells for several years, in a global market where installations remain heavily concentrated in a handful of countries. Construction has lagged because sites are unstructured, unrepeatable and wet. What has changed is that a small set of construction tasks — interior finishing, spraying, grinding, reinforcement fabrication, inspection and monitoring — are structured enough to automate profitably, and those are exactly the tasks on show.
The gap that remains is not technical. It is commercial and operational: who owns the machine, who maintains it, who trains the operator, who carries the risk if the finish fails inspection, and whether a Saudi contractor can get spare parts and a service engineer inside a day. Those are the questions being asked on the floor, and they are the same questions that decide whether any piece of industrial equipment sells in this market. Physical AI will scale in Saudi Arabia at the speed of its after-sales network, not at the speed of its models.