On the edge of the Indian Ocean, a Greek multinational is pitching one of the more unconventional infrastructure projects to emerge in Africa’s digital scramble: a $1.5 billion artificial intelligence data middle that would operate free of Kenya’s strained electricity grid.
Amaco Energy Group, based in Athens with operations in Houston, has proposed building a large-scale AI computing facility in Mombasa paired with an independent offshore power system fueled by liquefied natural gas. The company’s chief executive, Theodore Theodoropoulos, has been in Nairobi seeking government approvals for what it describes as a fully integrated, off-grid installation. Officials and the company frame the project as a response to East Africa’s rising appetite for cloud computing, digital finance, and AI workloads at a time when reliable power remains the region’s most stubborn bottleneck.
The price tag, Sh194 billion in Kenyan shillings or roughly $1.5 billion, would rank among the largest single private digital infrastructure bets contemplated in the region. Amaco’s concept centers on its Hercules platform: a floating offshore system that combines LNG-powered electricity generation, cooling infrastructure, and direct support for high-density AI servers. The company says the facility would not draw from Kenya’s national grid and could, in some scenarios, supply surplus power back into the network.
Kenya has positioned itself as East Africa’s emerging data-center hub. Nairobi already hosts the bulk of the country’s facilities, while Mombasa offers proximity to subsea cable landing stations that link the region to Europe, the Center East, and Asia. Demand is being driven by fintech, enterprise cloud adoption, and the early stages of local AI activity. Yet the continent’s broader computing footprint remains tiny by global standards, and power constraints have repeatedly complicated large projects.
That reality was underscored by the stalled Microsoft-G42 data middle initiative. In May 2024, Kenyan President William Ruto announced a landmark $1 billion partnership with Microsoft and the UAE-based AI firm G42 for a major facility near Olkaria in Nakuru County. The project was framed as transformational for East Africa’s digital economy. Progress later stalled. Kenya’s National Treasury withheld funding approval for the government’s portion, and the parties disagreed over Microsoft’s request for guaranteed uptake of cloud capacity. Power concerns loomed large as well. Kenya’s installed capacity stands at about 3,000 megawatts. The initial phase was planned at a manageable 100 megawatts, but the full buildout envisioned by G42 reached up to one gigawatt, equivalent to roughly a third of the country’s entire supply. By mid-2026 the project had been effectively suspended, illustrating how electricity availability, commercial terms, and fiscal caution can derail even well-backed hyperscale plans.
The Opportunities
If realized, the Amaco project could deliver several tangible benefits. First, it would add scarce high-density computing capacity tailored for AI workloads in a region that currently has limited AI-capable facilities. Kenya has only a handful of such centers compared with South Africa, and local processing could reduce latency and reliance on servers in Europe or the United States.
Second, the off-grid design directly confronts the central obstacle facing hyperscale AI infrastructure in Africa: grids that cannot absorb multi-hundred-megawatt loads without crowding out households and industry. An independent LNG-powered platform theoretically sidesteps that constraint and accelerates deployment, offering a potential alternative path after the Microsoft-G42 difficulties.
Third, the project could generate construction and specialized technical jobs, stimulate local supply chains, and reinforce Mombasa’s role as a connectivity and digital gateway. Amaco has suggested surplus electricity might feed the national grid, potentially easing power shortages in coastal Kenya. Broader regional effects could include attracting more cloud and AI investment, supporting data-localization preferences, and giving East African businesses and governments access to nearby compute resources for applications in finance, agriculture, health, and public services.
Finally, the model itself, pairing dedicated power generation with AI infrastructure offshore, addresses a global problem. Data centers already consume significant shares of electricity in advanced economies, and AI is accelerating that demand. Solutions that decouple compute growth from strained terrestrial grids may find buyers beyond East Africa. Amaco has outlined similar ambitions for the Center East and Europe.
The Risks
The proposal also carries substantial uncertainties and downsides. At the most basic level, it remains at an early stage: talks with Kenyan officials are underway, but no final approvals, financing details, or construction timeline have been publicly confirmed. The Microsoft-G42 experience shows that even high-profile, government-backed projects can unravel over power guarantees, offtake commitments, and fiscal assessments. Large infrastructure ventures in the region frequently face permitting delays, political shifts, and financing hurdles.
Energy and environmental risks are pronounced. LNG is a fossil fuel. While cleaner than coal or heavy fuel oil, its use locks in carbon emissions and exposes the project to global gas price volatility, supply disruptions, and evolving climate policy. Offshore platforms introduce marine environmental concerns, including potential impacts on fisheries, coastal ecosystems, and shipping lanes around Mombasa. Data centers also require significant cooling; even with integrated systems, water or thermal management in a tropical coastal setting will demand careful engineering.
Technical and operational risks are equally real. Floating or offshore AI facilities of this scale are largely unproven at commercial hyperscale. Reliability, cybersecurity of a maritime platform, maintenance logistics, and resilience to storms or security threats in the Western Indian Ocean all require rigorous demonstration. Any failure in the independent power system would idle extremely expensive computing hardware.
Economic and developmental questions linger. Will the facility primarily serve global AI training and inference workloads for foreign clients, or will it meaningfully expand access and capability for Kenyan and East African users? Data-protection implications are significant: a regional-scale AI middle processing personal, corporate, and government data from multiple countries would test Kenya’s Data Protection Act and raise complex cross-border transfer and jurisdiction issues.
There is also the risk of opportunity cost and dependency. Heavy reliance on imported LNG and foreign technology could limit local value capture. If surplus power materializes, it may help; if the platform remains fully isolated, Kenya’s broader electrification and industrial needs receive little direct benefit. Public perception already shows skepticism in some quarters about whether such projects primarily serve local populations or extract computational resources for distant markets.
Regulatory and political risks cannot be ignored. Approvals involving offshore energy, port-adjacent infrastructure, and data sovereignty will require coordination across energy, environment, maritime, and digital ministries. Changes in government priorities or disputes over fiscal terms could stall the project, as occurred with the Microsoft-G42 initiative.
A Test Case for Africa’s Digital Future
Amaco’s Mombasa proposal crystallizes the tension at the heart of Africa’s AI and cloud ambitions. The region needs far more computing capacity if it is to participate in, rather than merely consume, the next wave of digital transformation. Existing grids and conventional land-based models are struggling to keep pace, a lesson reinforced by the stalled $1 billion Microsoft-G42 deal. Creative engineering solutions that bring power and compute together offshore represent one possible path forward.
Yet creativity does not guarantee success. The project will be judged not only by whether the servers eventually hum, but by whether it delivers durable local economic value, manages environmental trade-offs responsibly, respects data-governance norms, and avoids becoming another high-profile announcement that fades into the permitting pipeline.
For Kenya and East Africa, the Amaco plan is both an opportunity to leapfrog infrastructure constraints and a high-stakes experiment in whether novel energy-compute hybrids can scale sustainably on the continent. The coming months of negotiations in Nairobi will determine whether the $1.5 billion vision moves from concept to concrete or remains an ambitious outline drawn on the waters off Mombasa.
Mohd Hassan has extensive experience in news gathering, editing, and writing for the newswire industry, Contact – Info@impactnews-wire.com
Burna Boy (Nigeria)
Most Famous Song: “Ye”
Short Bio: Damini Ebunoluwa Ogulu (born 1991), known professionally as Burna Boy, is a Nigerian singer,
songwriter, and record producer. He is one of the most prominent African artists of his generation, recognized for his distinct blend of Afro-fusion, reggae, dancehall, and R&B. He has achieved significant international success, including a Grammy Award.
Seydou Keïta (Mali)
Most Famous Photo: “Untitled” (often referring to his studio portraits from the 1950s, such as Untitled (Woman in
Headwrap with Flowers) or Untitled (Couple, Bamako)).
Short Bio: Born around 1921 in Bamako, Mali, Seydou Keïta is considered one of the most significant African studio
photographers of the 20th century. He opened his successful studio in 1948, capturing the evolving society of Bamako on the cusp of independence. His portraits, characterized by meticulous composition, attention to detail (including props and backdrops), and a dignified portrayal of his subjects, offer a unique visual archive of Malian life.
Spunky Comet
Fun South African Fact: The first successful heart transplant was in a Cape Town hospital
Houston Rocket
I thought I wanted a job, turns out I really just wanted a paycheck