The Chief Executive Officer of Hubtel, Alex Bram, has called for a national data-sharing framework that encourages companies to preserve and make data available under appropriate arrangements to support the development of artificial intelligence (AI) in Ghana.

He believes the absence of an effective system for sharing data between institutions remains a major obstacle to developing AI technologies locally, warning that investment in computing infrastructure alone will not be sufficient to drive Ghana's AI ambitions.

Mr. Bram wants the government to introduce incentives that encourage businesses to retain useful data rather than discard it, while establishing clear arrangements for how institutions can share information.

“At the least, we should have a data framework that will sort of call on companies to actually not delete data, but store them, and have a framework for incentivising companies to store data,” he said.

He explained that Ghana's planned national AI compute centre would require substantial amounts of data to support the development and training of AI systems.

“Because at some point when we finish the AI data centres, we will need a lot of data. But it's a step in the right direction,” he added.

Speaking in an interview ahead of the second episode of the Digital Economy Forum, Mr. Bram drew on Hubtel's experience developing its own artificial intelligence systems to explain why the country must prioritise data availability alongside investment in computing infrastructure.

The opportunities and challenges associated with AI adoption will form part of discussions at the second episode of the Digital Economy Forum, under the theme “AI and the Ghana Economy: Opportunity or Disruption?”

The forum, held in partnership with Hubtel, will feature the Minister for Communication, Digital Technology and Innovations, Sam George; Economist and Technical Advisor at the Finance Ministry, Dr Theo Acheampong; Founder of AI for Ghana, Adam Watkins; Head of Engineering at Hubtel, Augustine Adjei; and Vice President of IMANI Africa, Bright Simons.

It will air live on JoyNews and Joy 99.7 FM on Wednesday, October 21, 2026, at 8 p.m.

Hubtel's struggle to access data

Mr. Bram revealed that Hubtel established an AI laboratory in 2023 after beginning experiments with the technology in 2022.

The company initially sought to develop its own AI models using machine learning and neural network technologies, working with other businesses to test some of the systems it had developed.

However, the company encountered difficulties obtaining sufficient data to improve the performance of its models.

According to Mr. Bram, Hubtel approached several large companies to explore opportunities for sharing data but discovered that there were no clear arrangements governing how such exchanges should take place.

“At a point, we tried to connect with other large companies to see if they can give us, they can share data with us, we can share some of our data with them,” he explained.

“We realised there's still no framework in Ghana for sharing data. And so some of them were like, do we sign NDAs? Do we copy our data in Excel and give it to you? Do we have an agreement to share the data, and how do we even share it? There's just no framework.”

He said the uncertainty made companies reluctant to share information, limiting Hubtel's ability to obtain the data required to improve its AI models.

https://youtu.be/bfzoN-r37CE?t=3

“We just found that in Ghana it wasn't. So there weren't that many companies willing to share data with us to improve. So it became another obstacle that didn't really help our success,” he said.

Mr. Bram contrasted Ghana's situation with developments in the United States, China and parts of Europe, where he believes established data-sharing arrangements have contributed to the advancement of artificial intelligence.

He argued that Ghana must learn from these experiences by creating conditions that allow local technology companies to access useful data through structured arrangements.

Data critical to Ghana's AI compute centre

The Hubtel CEO welcomed the government's plans to establish a national AI compute centre, but maintained that the investment would achieve limited results without sufficient data to support the development of AI applications.

He described the proposed facility as an important step towards positioning Ghana in the global AI economy, while urging policymakers to accelerate the development of the supporting systems.

“So that data framework is important. If that AI compute centre is to succeed, because when a computer centre is there, without the data, it's nothing,” he stressed.

Asked whether the data-sharing framework should be established even before the physical computing infrastructure is completed, Mr. Bram agreed.

“Yes, yes. I think we need that. I think at the least, if not maybe enforcing it on private companies, I think at least we can enforce it with the government institution. Because AI is just data,” he explained.

His position suggests that government institutions could take an early lead in establishing clearer arrangements for retaining and sharing data, while incentives encourage private companies to participate.

Mr. Bram believes such a framework could have significantly improved Hubtel's own AI development efforts.

“We will need that. If we had that, I'm sure our AI lab would have been a success. But we didn't have that,” he said.

Protecting personal and commercially sensitive data

While expanding data availability could accelerate AI development, it also raises questions about how personal information and commercially sensitive records would be protected.

A national data-sharing framework would need to establish clear rules governing which information can be shared, who can access it, how it can be used and what safeguards are required to prevent misuse.

Such arrangements would be particularly important where information contains customers' personal details, financial transactions, confidential business records or other sensitive material.

The retention of data would also need to be distinguished from permission to share or use it for AI training. Preserving information does not automatically authorise its disclosure to third parties.