Saudi Arabia’s Data Centre Boom Is Creating a New HSE Challenge
Mission-critical construction changes the safety picture as projects move from civil works to MEP, energisation, commissioning and live operations.
Saudi Arabia is moving quickly from AI ambition to physical infrastructure. On 31 August 2026, HUMAIN and DataVolt announced that construction was underway at Oxagon, with 100 MW of the 360 MW first phase currently under development. Separately, Microsoft confirmed that its Saudi Arabia East datacenter region will become available to customers in November 2026, another indication of the Kingdom’s expanding cloud and AI infrastructure.
For teams delivering new facilities such as Oxagon, the bigger question is not the investment headline. It is what happens to the HSE risk profile when a project moves from concrete and steel into live electrical systems, integrated testing and 24/7 operations.
Why data-centre construction needs a different HSE lens
A data centre may look like another large building project, but its operational core is a dense network of electrical, mechanical and control systems. Switchgear, UPS systems, generators, batteries, cooling plant, fire systems, controls and communications infrastructure make MEP work a much larger part of the risk picture than on many conventional commercial builds.
The scale is also changing. The International Energy Agency notes that conventional data centres may sit around 10 to 25 MW, while hyperscale AI-focused facilities can be 100 MW or more. Higher compute density places greater demands on power and cooling, which carries through into installation, testing and commissioning.
The hazards themselves are familiar. What changes is the density of interfaces: multiple specialist contractors, high-voltage works, temporary power, heavy lifts, simultaneous operations, changing access rules and tight delivery programmes. The site can move from “under construction” to “partly live” in stages rather than through one clean handover.
The risk profile changes with the project
A static safety plan cannot fully reflect a project whose operating state changes week by week. The information that matters has to move with the work.
| Project phase | Typical HSE pressure points | What must stay visible |
|---|---|---|
| Civil works | Excavation, lifting, traffic, work at height, temporary power, heat exposure. | Temporary controls, exclusion zones, open actions. |
| MEP installation | Trade density, hot work, plant movement, cable and pipe routes, testing. | Package ownership, permits, repeated hazards, dependencies. |
| Energisation and commissioning | Live systems, switching, isolation, restricted access, integrated testing. | System status, authorised persons, access restrictions, defects and action owners. |
| Operations and handover | Maintenance in a live facility, contractor access, residual project works. | Residual risks, close-out evidence, lessons learned, operational records. |
Handoffs are where good HSE systems get tested
Take a fairly ordinary example. An electrician spots damaged temporary cabling outside an electrical room and sends a photo and voice note to a supervisor. The package contractor is told to fix it. A screenshot goes into a messaging group. Someone adds an action to a spreadsheet. Two days later, the commissioning team takes control of the room.
At that point the original hazard may have three records in three places, but no single story. Was the cable replaced? Who verified it? Did the access condition change when the room became controlled? Does the commissioning team even know the observation existed?
That is the practical information problem on subcontractor-heavy projects. One observation can move through a worker, supervisor, subcontractor HSE lead, principal contractor, client team and eventually operations. If the location, package, owner, deadline, evidence or final verification falls away during one of those handoffs, the issue has not really been closed. It has simply disappeared from somebody’s list.
Multilingual teams make clarity more important
Saudi data-centre projects bring together specialist firms and workforces from different language backgrounds. The challenge is not simply translation. It is whether a worker can report a risk accurately at the point of work and whether that meaning survives as the information moves through the organisation.
Voice, images and familiar reporting channels can help, but context still matters. “Cable on floor” is not the same record as “temporary power cable crossing the access route outside an electrical room.” During energisation and commissioning, location and system status can change the meaning of an observation within hours.
Saudi HSE requirements worth noting
Saudi Arabia’s HSE infrastructure is becoming more structured around reporting and investigation. The National Council for Occupational Safety and Health provides a service for reporting work-related accidents, injuries, occupational diseases and near-misses, with investigation procedures and documented outcomes.
For early civil works, climate also remains part of the risk picture. Saudi Arabia’s 2026 midday work ban applies to private-sector establishments from 15 June to 15 September, between 12:00 and 15:00, subject to specified exemptions. The broader lesson is useful: the dominant HSE concern changes as a programme moves from civil works to energisation and live-system interfaces.
What project leaders should review before the next phase
- Map the information flow. Know where observations, incidents and corrective actions are created, who owns them and where they are stored.
- Define the energisation boundary. Make changes in access, permits and system status obvious to every contractor affected.
- Use one close-out standard. “Closed” should mean completed, evidenced and verified, not simply removed from one contractor’s list.
- Make frontline reporting practical. Support the languages and formats workers can use accurately, including voice and images where appropriate.
- Carry HSE history into operations. Open actions, residual risks and recent lessons should not disappear when construction hands over to the live facility.
Where connected frontline HSE systems can help
There is no shortage of safety software on major projects. The harder question is whether the system shortens the distance between someone seeing a problem and somebody owning, fixing and verifying it. A polished dashboard sitting above disconnected spreadsheets and chat groups does not solve much.
This is where Navatech can play a practical role. Navatech’s nAI Flow captures safety inputs through the Navatech app and supported WhatsApp workflows, turning text, voice and images into structured HSE records and routing actions through to resolution. nAI Hub provides access to organisation-specific safety information through familiar interfaces, with multilingual and voice support.
For a Saudi data-centre programme, the value is not “AI for safety” as a slogan. It is continuity: keeping the original frontline signal connected to action ownership, close-out evidence and the project record as contractors and phases change.
Mission-critical safety depends on continuity
Saudi Arabia’s data-centre pipeline will keep creating technically dense projects with demanding handover programmes. The hazards are recognisable. What is harder to manage is the number of interfaces around them.
A project should be able to answer five questions about any meaningful safety issue: who raised it, where it happened, who owns it, what changed, and what proves it was resolved. If those answers survive the move from civil works to MEP, energisation, commissioning and operations, the HSE system is doing something useful.
Building or operating mission-critical infrastructure in Saudi Arabia?
Frequently asked questions
Data centres combine dense MEP systems, high-voltage infrastructure, heavy equipment and specialist contractors with a commissioning process that can overlap ongoing construction. Risk boundaries can change quickly as systems become live.
Because an observation may cross contractor, package and phase boundaries. Its owner, location, evidence and status need to remain visible until the action is verified, otherwise an issue can disappear from one register without being resolved for the project.
Relevant incident and near-miss history, open or recently closed actions, residual risks, temporary controls, energisation and access information, and lessons needed to understand the live facility safely.
