Over the last five years, artificial intelligence has shifted from a fringe interest to one of the most important drivers of global economic growth. So important has the technology become that the United Nations Security Council held its first open debate on artificial intelligence last month. While little of substance was achieved, a General Assembly resolution authorizing the creation of an independent scientific panel on AI may have a more enduring impact. One of the core questions this panel will seek to answer is how AI can support sustainable economic development without entrenching inequality.
The potential dangers here have deep historical parallels. AI runs on compute, cloud capacity, and data—resources that are concentrated in the hands of countries in the Global North. Africa, for example, hosts less than 1% of global data center capacity, leaving the continent reliant on expensive infrastructure abroad. Even an IT powerhouse like India hosts just 3% of global capacity, despite being home to nearly 20% of the world’s population. Meanwhile, workers across the Global South are earning as little as $2 an hour creating, cleaning, and labeling data for use in Western models.
A new digital colonialism?
To some, this looks like a digital version of the kind of resource extraction associated with the age of empires: labor and data flow inexorably north, where they create economic value, but little of this value finds its way back into the pockets of developing nations.
The reality is that these patterns are driven by market forces rather than imperial ideology, but the historical echoes are troubling nonetheless. Whatever the motivations, we know that this kind of concentration of power can do long-term economic and social damage. In some cases, the results are felt only in the underserved countries. AI systems trained to deliver healthcare to Western patients, for instance, can be dangerously inaccurate when working with other populations, limiting the transferability of the advances made in the West. Similarly, researchers at Columbia University have found that Large Language Models are less able to understand and represent the societal values of countries that have limited digital resources available in local languages.
These limitations are just the tip of the iceberg. AI is not just a productivity tool—it’s a force multiplier for innovation. It will shape how we farm, teach, heal, and govern in the future. If the Global South remains a passive consumer of imported AI systems, it risks losing not just economic opportunity but digital sovereignty. The Industrial Revolution brought extraordinary wealth to Europe and North America while locking much of the world into dependency for generations. AI could repeat that cycle—more rapidly and at an even greater scale.
Why this should worry every global business
The irony is that this approach hurts everyone, including the companies driving it. In terms of population, India has overtaken China while Nigeria and other African nations are enjoying booming birthrates. These countries represent tomorrow’s largest markets. Yet multinationals that treat them as data factories without trying to situate that data in its local context will find that they don’t understand the customers they will desperately need tomorrow. A model that misunderstands how most of the world thinks about family, risk, or trust is a model doomed to fail.
We have already seen how this trend can play out. The mobile money transfer company M-Pesa revolutionized banking in Kenya while Western banks were still trying to penetrate the market with credit cards. Today, Indian companies are developing chatbots that can speak to the hundreds of millions who communicate daily in so-called “low resource languages.” Unless multinationals begin to think intentionally about how they can serve these underserved populations, they will find themselves looking in from the outside once these markets mature.
The path forward
Avoiding the dangers of “algorithmic colonialism” and earning a position in emerging markets for AI products and services requires deliberate action from governments, businesses, and global institutions. Data centers, power supply, and research capacity should be financed like roads and ports, with blended capital from development banks and sovereign funds. Without local compute capacity, nations will inevitably remain digital renters, not owners.
Governments should also establish data trusts to negotiate how their citizens’ information trains global models, including setting benefit-sharing and transparency requirements. AI annotation work should pay living wages with proper labor protections. And critically, we need investment in open-source models, multilingual datasets, and local developers, so solutions are built with communities, not just for them.
Some companies are already changing course. They are investing in local infrastructure, creating genuine partnerships, and recognizing that sustainable profits come from creating value with communities, not extracting it from them. They understand that today’s data creators and workers will be tomorrow’s consumers, and, potentially, tomorrow’s innovators as well, if they are given the chance.
AI has the potential to be a great global equalizer—or it could become the most powerful driver of inequality in human history. We have seen what happens when transformative technology is hoarded: inequality deepens, resentment grows, and instability follows. If we want to write a different story—one in which the Global North and South cocreate the future and share the benefits of artificial intelligence—we must act now, before the gap becomes unbridgeable.
4 things leaders can do today to start bridging the AI divide
1. Audit your AI’s geographic blind spots today. Map where your training data comes from and which populations it represents. If more than 80% comes from Western sources, you run the risk of not being able to represent or communicate effectively with consumers from much of the world. Work to diversify your data if that is feasible, or develop localized AI systems that are trained or tuned with local data.
2. Create transparent data-sharing agreements. Develop a framework for using local data to train your models, including benefit-sharing provisions and audit rights for local data providers. Companies that move first will become preferred partners when governments start to mandate these arrangements.
3. Pay fair wages for AI work—and let your target markets know you are putting your money where your mouth is. Commit to paying local sustainable living wages plus a mark-up for data annotation and AI training work. Make this commitment public. You will attract better talent, improve the quality of your data, and build brand equity in emerging markets.
4. Launch an open-source initiative in at least one emerging market. Pick a specific challenge in a growth market—healthcare in Nigeria, agriculture in India, education in Indonesia—and commit to building an open-source solution with local developers. The relationships and market intelligence you gain will be worth more than any proprietary advantage you might give up.
