A subsea cable can land in Saudi Arabia long before a customer can buy a reliable connection from the coast to a data center. That gap matters as the Kingdom looks to host more of the cloud services and online content people use every day.
For Chief Carrier & Wholesale Officer at Salam, Amjad Arab, one measure of progress is how much of the internet traffic people use in Saudi Arabia is served from inside the Kingdom. Getting there depends on what happens inland. Cloud and content providers need places to connect, and customers need a second route that can keep their services running if the first one fails.
In a written interview with Inside Telecom, Arab discussed why the stretch from a cable landing station to a data center can be the hardest to deliver, how customers can tell whether two routes are truly independent, and what AI demand is already changing in Salam’s contracts. He also set out how the industry could measure Saudi Arabia’s progress in 2027 and 2030.
Saudi Arabia is positioning itself as more than a transit route between Asia, Africa, and Europe. In commercial terms, what would prove that the Kingdom has become a destination for hosting and exchanging digital traffic rather than simply carrying it through? Which measurable indicators would you prioritize: locally exchanged traffic, international capacity sold, cloud interconnections, enterprise workloads, or something else?
For me, the test is where Saudi traffic is served from. A transit hub carries content in from Europe or Asia and hands it over at the border. A destination hosts that content, caches it and exchanges it inside the Kingdom, so most requests are answered locally. If I had to choose one indicator, it would be the share of traffic consumed in Saudi Arabia that is delivered from inside Saudi Arabia. Next to it, I would put the number of cloud and content networks that interconnect inside the Kingdom.
International capacity sold still matters, but on its own it describes a corridor as easily as a destination. The supporting measures are cloud on-ramps and direct interconnections in the Kingdom, content delivery points of presence, and how fast high-value local services grow compared with pure transport.
The direction is already visible. Hyperscale cloud regions are opening in the Kingdom, in Jeddah, Riyadh, and the Eastern region, and data sovereignty rules keep more workloads local. In our own business, a growing share of new customers come to us to bring content and cloud into the Kingdom.
Salam has described its national fibre backbone and the ability to connect international landing stations with domestic data centres. Where is the most significant bottleneck today between a subsea cable landing and a customer-ready, end-to-end service: route diversity, interconnection, permitting, capacity economics, or the final connection into the data centre? What has Salam done to address that bottleneck?
In our experience, the hardest stretch is the inland segment between the landing station and the data centre. That is where several operators’ assets, permits, and commercial terms have to line up before a customer gets one service with one SLA. Capacity can be landed in Jeddah or Al Khobar well before a customer can buy a diverse, protected path into the data centre where its workloads sit. In practice, the gap shows up as route diversity on that inland leg and as the time it takes to agree interconnection across several parties.
Our answer has been to sell the whole path as one service. Cable landing station access in Al Khobar and Jeddah, our national fibre network, cloud backhaul and carrier interconnection now sit in a single wholesale offer, so a carrier or cloud provider deals with one partner from landing station to data centre, with more than one route option between the two. Internally, sales, engineering, delivery, legal and finance work complex orders jointly from day one, because the commercial structure and the technical design must be settled together if delivery times are going to fall.
Through its partnership with AFR-IX Telecom on the Medusa cable system, Salam has outlined a role in bringing new connectivity into Saudi Arabia through terrestrial backhaul and interconnection. As of late 2026, what is operational, what remains in development, and what milestones should carriers watch next? How will the resulting routes differ from existing options in practical terms?
Our role in Medusa is the Saudi side of the system. Under our agreement with AFR-IX Telecom, signed in October 2025, Salam is the gateway partner for Medusa in the Kingdom.
Medusa’s first segment, connecting Marseille and Bizerte, has been operational since June 2026. On the Saudi side, we are developing the infrastructure required to bring that connectivity into the Kingdom. We will interconnect with Medusa via Aqaba in Jordan, with the route extending through our terrestrial network and the interconnection point at Haql, and onward to landing stations, data centres and carrier points of presence across Saudi Arabia. The next priority is establishing the Aqaba Digital Hub POP, with the connection into Saudi Arabia targeted within 2027. Capacity swaps will further extend that reach into the wider Gulf.
The 2025 Red Sea cable disruptions showed how quickly faults can raise latency across multiple markets. How should wholesale customers distinguish genuine route independence from capacity that ultimately shares the same physical or geographic points of failure? Without disclosing sensitive network details, what service-level evidence should a carrier provide about automatic rerouting, restoration times, and performance during an outage?
Customers should ask where two routes actually separate. Two circuits can carry different names and different cable systems and still share a landing station, a conduit, a border crossing or a stretch of seabed. For a diversity claim to mean something, the provider has to show, at least at the level of landing stations, corridors and crossing points, where its paths are physically separate and where they meet.
On evidence, a carrier should be able to show a customer, under NDA where needed, three things: the availability each route has actually achieved, the latency on the protection path after a switch, and how often automatic rerouting is tested and what happened when it was used. Restoration objectives belong in the contract, with a clear target for each route. That is the standard we hold ourselves to with our customers.
Our own approach is to give customers paths that start in different places. Our landing stations sit on two coasts, Al Khobar on the Gulf and Jeddah on the Red Sea, and our terrestrial routes run north through Jordan towards Egypt and Europe, to northern Iraq, and east to the UAE and Bahrain. For business continuity, satellite capacity adds a further layer.
At Capacity Europe in October, you joined a discussion on fibre corridors for AI training, inference, and scale. In Salam’s wholesale business, what actual changes in customer demand can you identify so far: larger committed capacities, more diverse paths, tighter latency requirements, or new interconnection locations? Where is AI changing purchasing decisions today, and where is demand still more forecast than contracted reality?
The change we can already see in contracts is in the shape of demand more than its size. At Capacity Europe 2026, I participated in a keynote panel titled “Wiring the AI Boom: Building Fibre Corridors for Training, Inference and Scale.” Cloud and data-centre customers ask for paths from landing stations into specific facilities, for a second, diverse route as standard, and for interconnection in more locations inside the Kingdom. In our wholesale discussions, backhaul between cloud regions, landing stations and carrier hubs is now a regular item, and that links directly to cloud and AI workloads being hosted in Saudi Arabia.
AI training is where demand is still more forecast than contracted. The market is planning at very large scale: the National Data Center Strategy targets up to 1.5 GW of capacity by 2030, and SDAIA’s Hexagon facility in Riyadh alone is planned at 480 MW. Before committing, customers want to know the routes will be there. Most of the firm commitments we sign today, however, relate to cloud and inference platforms already in service. I expect training demand to reach the wholesale market in large, infrequent blocks once specific data-centre projects reach a firm build date. Our job is to have diverse routes ready before those decisions are made.
As AI applications move from large training clusters towards distributed inference and enterprise deployment, what changes in network design and commercial models? Is the stronger opportunity for operators selling bandwidth, low-latency access to compute, managed connectivity, or services that combine connectivity, security and cloud access?
In network terms, distributed inference is a lot like content delivery and gaming. Training needs very large pipes between a small number of sites. Inference needs compute close to users, low and predictable latency, and many interconnection points. Through CDN and gaming localisation, we have built a business around that profile, where every millisecond between the platform and the user shows up in the experience. As I said recently, the rapid expansion of hyperscalers and AI infrastructure is creating demand for high-capacity, low-latency and geographically diverse fibre networks, and these customers have a clear view of where their workloads are heading, often well before construction decisions are made.
For a wholesale operator, I think the stronger opportunity is selling a path to compute. Carriers, cloud platforms and content networks want low-latency, protected routes to the AI and cloud platforms hosted in the Kingdom, with performance commitments they can pass on to their own customers. Commercially, that moves us from long-term capacity sales towards services with latency and availability commitments attached. That is where the margin sits, and where a local operator with national fibre, landing stations and local interconnection adds something a global provider would find hard to build from outside.
Salam and Saudi Arabia’s Water Transmission Company announced a memorandum of understanding to explore fibre connectivity using complementary infrastructure assets. Based on your earlier experience in utility-linked telecom infrastructure, what conditions make this kind of cross-sector partnership commercially and technically viable? What concrete outcomes would show that the agreement has progressed beyond an exploratory framework?
Utility infrastructure can be a very efficient route for fibre, because the right of way, the corridor and much of the civil work already exist. In my experience, these partnerships work when three conditions are met. Both sides need a commercial reason that stands on its own. Operating responsibilities, from maintenance access to restoration after damage, have to be agreed in detail before anything is built. Finally, the regulatory position on using utility assets for telecom services must be clear from the start. In the Kingdom, that third condition is supported by an enabling regulatory framework: in 2023, CST and the Water Regulator signed an MoU aimed at maximising the use of water-sector infrastructure to support the expansion of fibre-optic coverage.
Our memorandum with the Water Transmission Company is exploratory, and we describe it that way. Progress will show when a commercial agreement replaces the memorandum and capacity on those routes is available for customers to buy. Until then, it remains an opportunity we are actively working on.
Hyperscalers and large content providers are increasingly shaping where traffic is generated, exchanged, and delivered. How does that shift affect wholesale carriers’ negotiating position and profitability? Where can a provider such as Salam retain value beyond price-per-gigabit competition, and how do you avoid becoming dependent on a small number of very large customers?
Hyperscalers and large content providers now decide where a large share of traffic is created and exchanged, and they buy with real scale and real expertise. Competing with hyperscalers only on price per gigabit, a carrier will lose margin every year. The relationship has also changed. A few years ago, operators saw hyperscalers mainly as competitors. Today the relationship is increasingly collaborative, because hyperscalers need local partners to reach customers in the Kingdom and to operate under Saudi data rules.
The value a local wholesale operator keeps is in what is hard to replicate: diverse physical routes into the Kingdom, the stretch from landing station to data centre, local interconnection, speed of delivery, and knowledge of the local market and its regulation. From a local partner, a hyperscaler needs exactly these, and we sell them as outcomes with commitments attached. The digital highways are being built; the commercial question is who collects the tolls.
On concentration, the answer is breadth. Our wholesale customer base has grown and now spans international carriers, cloud providers, content networks, gaming platforms and enterprises. Our revenue has diversified too: CDN, cloud and gaming services, which barely existed in our business in 2023, are now a meaningful line of their own. Although we work closely with the largest hyperscalers and content providers, our growth does not rest on any single one.
Saudi Arabia is expanding international gateways and domestic infrastructure, while regulators and operators are balancing investment incentives with market access. What would a genuinely open and competitive interconnection ecosystem look like for smaller carriers, content networks and new entrants? Which practical barriers still need to be reduced?
Salam relies on interconnection as much as it provides it, so we see openness from both sides. For us, an open interconnection market is one where a new carrier or content network can reach landing stations, data centres and exchange points on clear terms and within predictable timeframes. In practice, that means greater choice in landing and exchange facilities, transparent commercial terms for backhaul and cross-connects, and provisioning times that allow smaller players to plan their business with confidence. The goal should be to make it easier for a new entrant to connect without having to negotiate a separate arrangement at every point along the path.
The Kingdom has been taking steps in this direction. CST has established guidelines governing interconnection and the provision of landing-station and international cable-capacity services, while international connectivity continues to develop. SAIX also provides a neutral, non-profit exchange point designed to enable national and international networks to peer locally, reduce latency and support the localisation of traffic and content.
The next opportunity is to make the interconnection experience increasingly simple and predictable: easier access to facilities, efficient rights-of-way for new routes, and greater alignment of technical and commercial processes between operators. Common ordering processes and clearer lead times would help further reduce friction. When a global content network finds Saudi Arabia as straightforward to connect to as the international hubs it already uses, the case for exchanging more traffic locally becomes stronger. That is also the principle behind Salam’s focus on carrier-neutral interconnection.
Looking towards 2027 and 2030, what are the two or three measurable outcomes by which the industry should judge the success of Saudi Arabia’s wholesale connectivity strategy? And what is the most underestimated risk that could prevent infrastructure investment from translating into reliable services and sustainable economic value?
By 2027, I would judge progress on two measures. The first is the share of the Kingdom’s own traffic served from inside the Kingdom, ideally published as a national indicator that the whole industry reports against, alongside the measures CST already tracks. The second is the number of international routes into Saudi Arabia that are physically diverse, with their own landing points and inland paths. By 2030, I would add the volume of cloud, AI and content workloads that international customers host in Saudi Arabia to serve the wider region, because that is how the Kingdom earns revenue as a destination.
The risk I think is most underestimated, in any market building at this pace, is the time it takes to turn capacity into services customers can use. New cables and data centers are announced and counted. Yet customers buy services, and services depend on diverse inland routes, efficient interconnection between operators, and Saudi engineers with the skills to run those networks and restore them quickly. The physical foundation is in place. The service layer, and the professionals who run it, now must grow at the same pace.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Interviews to stay informed and up-to-date with our daily articles.