The rise of artificial intelligence has been heralded as a transformative moment in business, offering new efficiencies and innovation at a scale not seen in decades. With minimal infrastructure required to adopt many modern AI tools, one might expect them to spread rapidly across the globe, leveling the playing field for firms regardless of geography or size.
However, the global adoption of AI remains highly uneven. While firms in the United States and other English-speaking countries have quickly embraced tools like AI-powered chatbots and recommendation engines, many companies across Asia, Latin America, and non-English-speaking Europe lag behind.
New research by Harvard Business School doctoral candidate Dafna Bearson and Columbia Business School Assistant Professor of Business Nataliya Wright investigates this disparity. Their research identifies a surprisingly overlooked factor: the strategic market choices made by AI entrepreneurs themselves. Rather than developing tools for a truly global audience, many AI startups concentrate their efforts on English-speaking markets, where investors, customers, and user feedback loops are perceived to be most valuable.
"Managers today are overloaded with information, so they don't have time to look into every tool out there. They're going to look at the tools that are sent to them, or targeted at them," Wright said.
The result is a global digital divide that persists not because of access or skill gaps, but because companies outside the “core” markets aren’t being targeted in the first place.
How the Research Was Done
Bearson and Wright compiled a dataset covering 88,000 business-to-consumer firms across 171 countries between 2012 and 2023. They identified whether firms adopted AI tools by analyzing their websites with data from BuiltWith, a service that detects technologies embedded in web platforms. These tools ranged from customer-facing chatbots to natural language processing engines and analytics dashboards.
They then assessed which companies were targeted by examining whether they were headquartered in English-speaking countries or used English-language tools on their websites. Bearson and Wright reasoned that AI entrepreneurs who created English-language products — most of whom also operated English-only websites — primarily targeted firms in markets like the U.S., UK, Canada, and Australia.
The study was also informed by qualitative interviews with AI startup founders in multiple countries. These conversations helped the researchers understand the business incentives behind entrepreneurs' geographic targeting decisions.
What the Researchers Found
Despite the widespread accessibility and ease-of-use of many AI tools, adoption remained surprisingly low — just 14 percent of firms in the dataset implemented any AI technologies during the study period. Adoption was significantly higher among firms headquartered in English-speaking countries. Companies in those countries were 12 percent more likely to adopt AI. The gap did not narrow when assessing firms who already installed English-language tools and therefore were more likely to see an appropriate use case for the AI technology, suggesting that “fit” frictions were not driving the gap in the first place.
“Technologies like AI can be important drivers of firm performance and also innovation, but the gains associated with these technologies might largely reach the hub regions because of entrepreneurs’ targeting decisions,” Bearson said.
The research also showed that the majority of AI entrepreneurs designed their websites in English and tailored their products to English-speaking users. This disproportionately funneled AI resources and exposure toward a narrow band of firms, while creating information gaps for companies elsewhere. Businesses that these entrepreneurs did not target faced what the authors termed “search frictions” — they simply weren’t seeing or hearing about these tools, even when the tools were technically adaptable.
One particularly important insight emerged when the researchers examined the performance outcomes of firms that had adopted AI. Companies located in non-targeted markets saw no weaker, and even stronger gains than their peers after adopting AI — a finding that directly challenges the assumption that AI is less useful in peripheral regions.
In interviews, AI startup founders frequently cited investor expectations and the perceived advantages of entering large, established markets as motivations for targeting English-speaking firms. One founder from Australia explained that winning contracts with major U.S. retailers was not just a business win, but also created a data advantage that made their product more defensible. A founder from Israel noted that their investors explicitly advised against prioritizing smaller, non-core markets, reflecting an industry-wide belief that success must begin with penetration into the US market.
Why the Research Matters
For business leaders and policymakers, these findings challenge the idea that frontier technologies naturally spread across borders. Even when tools are technically easy to implement, adoption is shaped by upstream strategic decisions, especially by startups building those tools.
This research also has implications for economic development and technology policy. If AI tools deliver outsized benefits to firms in emerging markets, then targeted incentives — such as subsidies, tax credits, or grants — could encourage entrepreneurs to build products for those regions. Policymakers might also consider tying government funding to efforts that promote adoption outside English-speaking hubs.
For AI startups, the results reveal a significant opportunity to expand beyond saturated markets. While English-speaking countries remain lucrative, they are also competitive and heavily targeted. Firms in non-core markets may represent an underexplored growth opportunity — and, according to the study’s findings, may actually generate stronger returns on AI adoption.
And for businesses in under-targeted regions, the message is clear: don’t assume AI tools aren’t relevant to you. The problem may not be the technology — it may be that no one’s tried to sell it to you yet.
FAQs:
Q: Why is AI adoption lower in non-English-speaking countries?
A: The study shows it’s not about access or technical capacity. Instead, many AI tools are developed with English-speaking markets in mind, meaning firms elsewhere don’t see them.
Q: Do companies outside the U.S. benefit less from AI?
A: No, in fact firms in non-English-speaking countries that adopted AI experienced greater funding gains on average than those in core markets.
Q: Why would entrepreneurs limit their audience to just English-speaking markets?
A: Because of investor expectations, perceived scalability, and easier access to large enterprise customers. Many AI startups are encouraged to prove traction in the U.S. or UK before expanding elsewhere.
Q: What can companies or governments do to address this imbalance?
A: Companies can invest in the proactive discovery of AI tools and local adaptation strategies. Governments might consider supporting AI tool localization or incentivizing startups that serve underrepresented regions.
Adapted from “Strategic Targeting and Unequal Global Adoption of Artificial Intelligence” by Dafna Bearson of Harvard Business School and Nataliya Wright of Columbia Business School.
