We may have spent two decades trying to solve the problem of internet access, only to be creating a new challenge: AI access. Only around 3% of US households currently pay for any AI service, according to the Bank of America Institute, even as headlines suggest generative AI adoption is already universal. In fact, the next digital divide may be between AI hype and actual usage. Telecom operators, who built much of the infrastructure to narrow the last digital divide, have a stake in how this gap narrows too.
We can think of AI access as a ladder of needs, with each rung dependent on the one below it:
- Literacy to use it effectively
- The quality and integration of that access
- Access to a useful AI assistant
- A capable device
- Connectivity
In many places, the bottom rung is subsidised for low-income households. Do we need to subsidise access to the tools higher up the ladder? If society eventually decides that it does, operators will almost certainly become part of that delivery infrastructure, in the same way they already are for broadband.
The Great Divide
Having said that, the digital divide still exists in many places. The ITU’s Facts and Figures 2025 report counts 2.2 billion people offline globally, with internet use at 94% in high-income countries against 23% in low-income ones. Any conversation about an AI divide sits on top of an intractable connectivity divide.
In Q1 2026, Microsoft’s AI Economy Institute found that 27.5% of working-age people in developed economies had used a generative AI tool, versus 15.4% in developing ones. That gap widened by 1.5 percentage points in six months, even as global adoption reached 17.8% of the working-age population. The factors underlying the digital divide, such as access to connectivity, digital skills and, in some regions, electricity, underpin a divide in AI access and usage too.
Even within wealthy countries, income predicts AI usage almost as accurately as it does between rich and poor ones. A UK-focused study found regular ChatGPT use is roughly twice as common among higher earners as lower earners. Choice of tool also appears to be dividing along income lines. In the US, Epoch AI analysed Ipsos survey data and found assistant use varies by income: higher earners skew toward Claude, lower earners to Meta’s AI.
Affordability seems to be a key driver. A median spend of $20 a month is already out of reach for a meaningful share of households – hence the Bank of America’s statistic that only 3% of US households pay. Economists Alex Imas and Soumitra Shukla put this in a global context: a premium ChatGPT subscription equals roughly 39 months of income in some countries.
OpenAI’s own research with Harvard economist David Deming found ChatGPT’s demographic gaps have narrowed as its user base has broadened. Gender gaps have narrowed and usage has grown faster in low and middle-income countries over the past year than in wealthy ones. While at first glance this sounds contrary to the other statistics cited, we should see this as growth taking place from relatively low bases in less wealthy countries, as opposed to market saturation in the wealthier ones.
Closing the Gap
Governments have started to respond to this gap.
India’s Yuva AI for All is a free training course targeting ten million citizens with AI literacy, but it teaches skills rather than granting tool access. Singapore’s 2026 budget goes further on paper, offering six months of free premium AI access to citizens completing selected courses from the second half of the year. Estonia’s AI Leap programme already gives around 20,000 upper-secondary students and 3,000 teachers free AI tool access and training, expanding to 38,000 more students in September 2026.
What still appears to be missing is a policy framework that distinguishes between universal access to free AI tools and targeted support for low-income households.
In many countries operators already run the administrative machinery behind the one AI-adjacent subsidy that widely exists, which is subsidised broadband access for people who would not otherwise be able to afford it. Eligibility verification, billing integration, and distribution for social broadband and mobile tariffs often sit with the network operator. It would be reasonable to expect future AI-access entitlements to sit on that same infrastructure.
Verifying who qualifies for a subsidy, at scale, without excluding the people it is meant to reach, is a difficult operational problem, but one which operators already solve for broadband. Identity verification, income-tier eligibility checks and fraud prevention around subsidised access are capabilities built over years, not something. An AI vendor would struggle to stand up something like that quickly.
There is also a distribution role. If a government or employer eventually wants to bundle AI access and training, as per Singapore’s example, operators are natural distribution partners. Their billing and operations systems are thoroughly used to bundling different types of service. That puts operators in a good position to negotiate commercial terms with AI providers.
The Telco in AI Policy
Operators also occupy a more neutral position than most other participants in the AI value chain. Their commercial interest lies in users having the connectivity and device capability to use AI tools, rather than in which specific assistant wins more market share. This gives operators a useful perspective to bring into a policy conversation currently dominated by the AI vendors.
None of this determines what a fair AI access policy should look like, or even whether we certainly need one. However, the gap it would need to close is real today. The next twelve months of data should show whether AI access converges the way mobile broadband eventually did or whether, without intervention, it settles into a two-tier structure. For now, this is an emerging trend worth tracking.
Image by Vilius Kukanauskas from Pixabay
