Kenya just took a meaningful step in the race for AI-ready infrastructure. Safaricom has signed a strategic deal with iXAfrica Data Centres to deliver enterprise-grade services from NBOX1, a new Nairobi campus designed for high-power, high-density computing. Phase one of NBOX1 went live in February 2025 with 780 racks and 4.5MW of IT capacity on a site planned to scale to 22.5MW. iXAfrica is marketing it as East Africa’s first AI-optimized build, aimed at the kind of workloads that need serious power and low latency.
The partnership is simple on paper but big in impact: iXAfrica brings the facility and scale; Safaricom brings the network, enterprise relationships, and a packaged stack of services. Together, they’re targeting customers across East Africa—banks, telcos, fintechs, healthcare providers, logistics platforms, public agencies—who want to run sensitive, compute-heavy tasks inside the country without compromising speed or compliance.
Under the plan, customers can buy dedicated enterprise and cloud suites starting at 350kW with the option to scale past 1MW. That’s the kind of capacity you need when you’re training models, running large analytics pipelines, or deploying GPU clusters. The partners are also bundling integrated connectivity so companies get one path to compute, storage, network, security, and support—rather than stitching it all together themselves.
Why this partnership matters
AI is no longer a side project for big firms; it’s moving into core operations. That shift demands reliable power, efficient cooling, tight security, and very fast, very stable connectivity. Hosting this inside Kenya helps with data residency and performance. Safaricom CEO Dr Peter Ndegwa framed it as a direct response to that demand, saying businesses now need robust, scalable, secure infrastructure that can handle the next wave of innovation, including AI workloads, while keeping data sovereign and compliant.
Local hosting also reduces latency for users in Nairobi, Mombasa, Kampala, Dar es Salaam, and beyond. It helps with regulatory requirements too. Kenya’s data protection regime has pushed more enterprises to keep certain data in-country or under stricter controls, and critical sectors like finance and public services have little room for error. Placing AI-heavy operations inside a facility built for high-density racks—and backed by a major carrier—can simplify audits and improve user experience at the same time.
iXAfrica Chairman Guy Willner cast the deal as a pairing of strengths: AI-ready infrastructure on one side and Safaricom’s market reach and customer trust on the other. That combination makes it easier for enterprises to move from pilot to production. Instead of hunting for GPU-ready space, cross-campus connectivity, and 24/7 managed support from different vendors, they can buy a complete package with clear service levels and a single point of accountability.
Here’s what’s on offer under the agreement:
- Joint go-to-market strategy focused on enterprise and cloud customers across East Africa.
- Dedicated suites starting at 350kW, with scalability beyond 1MW for AI, analytics, and high-density compute.
- Integrated connectivity that leans on both firms’ networks for low-latency, resilient links.
- Safaricom’s end-to-end stack: colocation services (2U+), cloud, managed services, cybersecurity, and business resilience.
If you’re wondering who needs 350kW and up, think about a bank rolling out real-time fraud detection using GPU inference, a health-tech provider doing heavy imaging analysis, an e-commerce platform training recommendation models, or a broadcaster moving to 4K/8K workflows and live transcoding. These are workloads that don’t fit neatly in small rooms. They need predictable power and cooling, minimal jitter, and room to grow fast without redesigning everything every quarter.
NBOX1 is built for that. iXAfrica has positioned it as an AI-optimized campus, which typically means support for higher power densities per rack, advanced cooling options, and tight integration with carrier networks and cloud on-ramps. It’s also designed to scale. With 4.5MW live and a path to 22.5MW on-site, operators can carve out space for larger suites without moving customers between buildings. That matters when you want to add hundreds of kilowatts for a new cluster in months, not years.
Safaricom’s role is to make the whole thing usable day one. Its enterprise portfolio stretches from basic colocation (as little as 2U) to managed firewalls, DDoS protection, hybrid cloud design, backup and disaster recovery, and ongoing monitoring. That’s attractive for teams that don’t want to staff a big infrastructure unit just to support an AI rollout. The result is a single contract that covers facility, connectivity, and operations—something procurement officers tend to like.
What the rollout looks like
NBOX1 is only the start. iXAfrica has two more builds in motion: a planned NBOX1.2 phase with 18MW of IT capacity and a larger 11-acre NBOX2 campus at Tilisi slated for 53MW. If both move ahead as designed, Nairobi will have a clear runway for growth—enough to anchor regional AI deployments, multi-cloud strategies, public-sector digitization, and regional content delivery.
That capacity could mean a few things on the ground. First, shorter timelines for enterprises to migrate from overseas hosting to local high-density space. Second, more room for cloud-adjacent services—edge nodes, GPU-as-a-service, and low-latency interconnects between telcos, fintechs, and content platforms. Third, a boost for the local vendor ecosystem: electrical and mechanical contractors, cybersecurity firms, systems integrators, and managed service providers all have more to do when a campus scales.
There are challenges to watch. Power costs and reliability always matter in data centre economics. Operators need to drive efficiency, manage cooling carefully, and secure diverse power feeds. Talent is another factor. Running AI workloads at scale requires engineers who understand not just networking and virtualization, but GPUs, model serving, and data pipelines. The partnership helps by abstracting some complexity, yet the skills gap won’t fix itself overnight.
Connectivity is less of a bottleneck than it used to be. Kenya benefits from multiple subsea cables landing at the coast and growing terrestrial fiber across the region. With integrated routes from Safaricom and interconnect options at the campus, enterprises can stitch together multi-site redundancy and keep latency predictable. That’s key for financial trading windows, real-time logistics, telemedicine consults, and AI inference at the edge.
For CIOs, the practical checklist is straightforward:
- Map which workloads must stay in-country for compliance and performance, and which can sit in public cloud.
- Estimate near-term power needs (think in 50–100kW blocks) and plan for burst scenarios.
- Decide on network topology and redundancy before moving a single rack.
- Bake in security and observability from day one—identity, encryption, logging, and incident response.
- Negotiate clear SLAs covering uptime, response times, and capacity expansions.
Safaricom and iXAfrica are pitching this as more than capacity. It’s a bridge to advanced workloads—AI training, inference, data analytics, edge computing, and disaster recovery—hosted locally with enterprise-grade controls. If they execute as planned, Nairobi becomes a go-to node for East Africa’s next wave of digital services.
The signal here is hard to miss: East Africa wants to compete in the global digital economy on its own terms. With NBOX1 live, an 18MW expansion on the books, and a 53MW campus in the pipeline, the region is setting the stage for bigger bets in AI, cloud, and data-heavy applications. For businesses ready to scale, the arrival of a true hyperscale data centre option in Nairobi changes what’s possible—and how fast they can move.
joy mukherjee
September 12, 2025 AT 19:00Seeing the rollout of NBOX1 really underscores how quickly East Africa is catching up with global AI infrastructure trends. The partnership gives local firms a chance to keep their data close to home, which eases compliance worries and cuts latency for users in Nairobi and beyond 😊. It also means banks can run fraud‑detection models on‑premise instead of sending sensitive transaction data across oceans. Healthcare providers will finally have the horsepower to process high‑resolution imaging without waiting for distant clouds. Logistics companies can crunch route‑optimization algorithms in real time, sparing precious minutes that add up over hundreds of shipments. By bundling connectivity, power, cooling and security, Safaricom and iXAfrica remove a lot of the integration headaches that smaller teams usually face. The modular power options, starting at 350 kW, let businesses scale organically as their AI workloads grow. With a path to 22.5 MW on site, even the most data‑hungry AI labs can find room without having to relocate later. The focus on high‑density racks and advanced cooling also signals that energy efficiency is being taken seriously, which is crucial given the region’s power cost concerns. Talent development will be key, and the joint offering could spur more training programs for GPU‑focused engineers. Local contractors and service providers will see a boost in demand for specialized cooling and power‑distribution work. The combined brand trust of Safaricom and the technical depth of iXAfrica should give customers confidence in service‑level guarantees. From a regulatory standpoint, keeping data within Kenya aligns with the data protection act and helps avoid cross‑border complications. Moreover, the regional fiber backbone and subsea cables mean the campus can serve neighboring markets like Uganda and Tanzania with low latency. In short, this isn’t just another data centre – it’s a strategic platform that can accelerate AI adoption across multiple sectors in East Africa. 🌍
Rob Chapman
September 15, 2025 AT 02:34The thing about these kinds of deals is they open doors for folks who thought AI was only for the big tech giants. By having a local hub, startups can test ideas without massive upfront spend. It also brings a sense of community when you see different industries sharing the same infrastructure. Think of it as a digital co‑working space for heavy compute, where everyone benefits from the shared resources. It’s a win‑win for the ecosystem.
Delaney Lynch
September 17, 2025 AT 10:07I love how this initiative not only provides massive compute power, but also integrates security, networking, and managed services, which means companies don’t have to juggle multiple vendors, they can focus on building smarter models, and they get a single point of accountability, something that’s often missing in complex deployments, plus the scalability from 350 kW up to megawatt levels offers a growth path that aligns with budget cycles, and the fact that the data stays in‑country satisfies regulatory needs, creating a win for compliance teams as well.
Nicholas Mangraviti
September 19, 2025 AT 17:40Exactly, local access trims latency and cuts costs dramatically.
Jared Greenwood
September 22, 2025 AT 01:14Let’s cut the fluff-East Africa is finally getting the muscle it deserves, and this campus is the backbone for sovereign AI workloads, with power density, tier‑four uptime, and carrier‑grade redundancy all baked in, so any foreign provider thinking they can out‑compete on price will discover that local compliance, latency, and talent pipelines give us a decisive edge.
Sally Sparrow
September 24, 2025 AT 08:47Honestly, the hype around this new data centre feels a bit overblown; while the specs sound impressive, the real test will be whether Kenyan firms can afford the 350 kW entry point and whether the power grid can sustain the promised 22.5 MW without frequent brownouts. Too many projects promise miracles and then stall because of hidden OPEX, so I’ll be watching the actual utilization numbers before I get excited.