Bright Simons questions viability of $270m GhanaChat project

Bright Simons, Vice-President at the IMANI Centre for Policy and Education, has raised serious concerns regarding the technical and financial feasibility of Ghana’s national Artificial Intelligence (AI) strategy. Specifically, Simons highlighted significant mathematical errors in the government's computing capacity targets and questioned the US$270 million price tag for the "GhanaChat" project. According to Simons, the current roadmap for a national AI data centre lacks fiscal sustainability and relies on flawed technical benchmarks that could undermine the country's technological ambitions.

A primary point of contention involves the computing power targets set for the years 2027 and 2035. Simons pointed out that the strategy documents specify targets of "1021 FLOPS" and "1025 FLOPS," which appear to be typographical or conceptual errors intended to represent exponential values. Taken literally as written in the strategy, these figures would reduce the nation's total computing capacity to that of a desktop computer from the 1970s. Simons noted that even if the intended higher targets were achieved, the proposed data centre would only meet approximately 1% of the 2027 requirements, rendering the strategic planning fundamentally inconsistent.

Beyond the technical discrepancies, Simons warned of a mounting financial burden on the state. He estimated that the operational costs of the US$270 million data centre would likely outstrip its revenue generation, potentially requiring an annual government subsidy of US$61 million to stay afloat. He questioned the logic of investing such massive sums into a national platform designed to compete with established global AI giants, suggesting that the project as currently envisioned represents a significant fiscal risk to the national treasury.

As a more manageable and cost-effective alternative, the IMANI Vice-President proposed the development of a "sovereign AI assistant" specifically tailored for the civil service. This alternative project would require an initial investment of approximately US$13 million and rely on 144 Graphics Processing Units (GPUs)—a fraction of the cost of the GhanaChat initiative. Simons urged the government to conduct a thorough critical review of the national AI policy to prevent costly failures and ensure that public funds are directed toward realistic, high-impact technological improvements.