Gbenga Adegbiji runs Geniserve, a Nigerian digital infrastructure builder. He told an Africa Hyperscalers webinar that Nigerian data center operators out-engineer their American counterparts on power resilience, according to Techeconomy's coverage. His claim: "You possibly have people that understand how to engineer resilience in power in Nigeria much more than you have in the US." The panel convened around Nigeria's Central Bank directive, which forces banks and fintechs to store payment data locally by January 1, 2027. Adegbiji used the moment to argue that Nigeria's power problem produced better engineers.
Nigeria's national grid generated as little as 3,940 megawatts in March 2026 and hovered near 4,140 megawatts in January, against demand from over 200 million people. The grid collapsed at least 16 times across 2024 and 2025, based on outage tracking from Nigerian outlets including Guardian Nigeria and Pulse Nigeria. Gas shortages compound the shortfall: thermal plants need roughly 1,588 million standard cubic feet of gas daily and have been getting closer to 652 million. Adegbiji's argument: operators who build inside that reality treat generators, redundancy, and disaster recovery as day-one baseline requirements. Engineers in grid-stable markets, he said, still size backup power for hours. Nigerian operators size for disruption measured in days.
Ayotunde Coker, CEO of Open Access Data Centres, backed the certification point on the same panel. Adegbiji put it directly: "There is no Nigerian Tier 3 or UK Tier 3. It is Tier 3." Tier ratings, defined by Uptime Institute, test facilities for power, heat rejection, capacity, and operational reliability regardless of country. That standard is where grid-stressed markets tend to prove themselves: a facility that has survived actual collapses gets tested by conditions a lab can only simulate. The same dynamic shows up wherever grid constraints shape data center development, from Lagos to the US interconnection queue.
No one on the panel addressed cooling directly, but the resilience model carries a thermal cost. Operators who run generators as continuous primary power are managing two heat sources at once. IT load produces its usual heat. The generator engines producing that power reject heat of their own, through radiators and jacket-water systems sized for continuous duty rather than an occasional test cycle. That doubling lands on Lagos ambient conditions already in the high 80s Fahrenheit with heavy humidity. The wet-bulb environment makes heat rejection harder than in temperate, grid-stable markets. The same math is arriving in US markets building off-grid gas plants to power AI campuses. Once generation moves onsite and runs near-continuously, its waste heat becomes the cooling plant's problem too.
Operators who have already solved for that size chiller and cooling tower capacity around generator waste heat as well as IT load. Some are exploring absorption chillers that turn onsite generation's waste heat into usable cooling capacity. Adegbiji made an engineering-culture argument. The thermal accounting is ours to add: the two problems share one rejection loop. Any campus running a generator as primary power has to size that rejection loop for the engine as well as the rack. That holds whether it sits in Lagos or backs a Texas gas-plant deal.