The global rush to build artificial intelligence may be making the basic gateway to the internet more expensive for the people who need it most.
That is the uncomfortable warning from the GSMA’s State of Mobile Internet Connectivity Report 2026, released in Hong Kong on September 18. The industry body, which represents mobile operators and the wider mobile ecosystem, says more than 3.4 billion people still do not use mobile internet even though over 90 per cent of them already live within mobile broadband coverage.
In plain terms, the networks are largely there. The devices are not.
Also Read: Policy warning: Without intervention, AI could deepen the digital divide
The report argues that the next digital divide will not simply be between countries with advanced AI models and those without them. It will be between people who can afford the basic tools needed to access AI-enabled services, and those who cannot. For low- and middle-income countries, including several large Southeast Asian markets, that distinction matters. AI in healthcare, education, farming, financial services or public administration means little if the intended users cannot get online in the first place.
“Artificial intelligence has the potential to improve lives on an unprecedented scale, but AI is meaningless if people cannot get online in the first place,” said Vivek Badrinath, Director General of the GSMA.
The smartphone bottleneck
The report’s central point is simple: handset affordability is now the biggest barrier to mobile internet adoption across surveyed low- and middle-income countries, ahead of even digital skills.
By the end of 2025, an entry-level internet-enabled handset cost the poorest 20 per cent of people in low- and middle-income countries the equivalent of 44 per cent of their average monthly income. In Sub-Saharan Africa, that figure rose to 76 per cent.
Those numbers are not abstract. For a low-income worker, a basic smartphone can represent a choice between connectivity and household essentials. In Southeast Asia, where mobile phones are the primary internet device for many users, the cost of entry-level handsets directly affects whether people can access digital payments, government services, online learning, telemedicine and job platforms.
The GSMA says 4.8 billion people now use mobile internet on their own device. But growth is slowing. Around 160 million people came online in 2025, down from 190 million the year before. Meanwhile, 3.1 billion people live within mobile broadband coverage but do not use mobile internet, what the GSMA calls the “usage gap”. Most of them still do not own an internet-enabled device.
This is the gap that policymakers in emerging Asia have struggled with for years. Extending 4G coverage to rural islands, mountain communities or secondary towns is difficult but measurable. Getting affordable devices into people’s hands is harder, especially when household incomes are under pressure and device prices begin to rise.
How AI is pushing up phone costs
The new pressure point is the global AI infrastructure boom. Demand for data centres, servers and AI chips has increased competition for memory and other components also used in smartphones. According to the GSMA report and data from Counterpoint Research, memory prices more than doubled between the third quarter of 2025 and the first quarter of 2026. They then rose by a further 80 to 90 per cent in the second quarter of 2026.
The effect is already showing up in the handset market. Entry-level smartphone prices are rising sharply, while global smartphone shipments are forecast to suffer their largest annual decline on record. The report says the fall is driven mainly by the collapse of the sub-US$100 handset segment, with emerging markets expected to be hit hardest.
Also Read: The digital divide: Islands of modernity in a K-shaped economy
That creates a strange contradiction. AI is being promoted as a tool to widen access to knowledge, improve public services and boost productivity. Yet the same infrastructure race powering AI, one that analysts had already flagged could strain the wider semiconductor supply chain, is pushing up the cost of the devices many people need to use those services.
The GSMA says that until a year ago, reducing the price of entry-level smartphones to US$30 could have made devices affordable for almost 1.6 billion people. Reaching US$20 could have brought affordability within reach for around 2.2 billion people living under mobile broadband coverage. Those price points are now out of reach, the report warns, despite efforts by operators and manufacturers, including the GSMA’s Handset Affordability Coalition.
The body is urging chipset and memory manufacturers to increase the availability of affordable components for entry-level handsets. It is also calling for dialogue among component suppliers, mobile operators, device makers, policymakers and multilateral financial institutions.
That is a broad appeal, but the logic is sound. If AI-driven demand continues to absorb component supply at the high end, the bottom of the smartphone market could be squeezed further. For Southeast Asian countries still trying to close rural, gender and income-based digital gaps, that would be a serious setback.
Why Southeast Asia should pay attention
Southeast Asia is often discussed as a mobile-first region, but that phrase can hide uneven realities. Singapore has near-universal connectivity and is already positioning itself around AI governance, data infrastructure and digital public services. Indonesia, the Philippines, Vietnam, Cambodia, Laos and Myanmar face a more complicated picture, with large populations outside major cities still constrained by affordability, skills or service relevance.
The GSMA’s warning is therefore directly relevant to the region. Many Southeast Asian governments are building digital identity systems, e-payment networks, online tax platforms, telehealth tools and AI-assisted public services. Startups are doing the same in credit scoring, education, logistics, agriculture and small-business software. But these models often assume that the user already owns a capable smartphone and can afford data.
If entry-level devices become more expensive, the business case for inclusive digital services weakens. A farmer cannot use an AI crop advisory tool without a device. A micro-merchant cannot adopt digital bookkeeping without reliable mobile access. A student cannot benefit from personalised learning apps if the household shares one outdated phone.
The economic stakes are significant. Previous GSMA analysis estimates that closing the mobile usage gap would generate US$3.5 trillion in additional GDP between 2023 and 2030, with more than 90 per cent of those benefits flowing to low- and middle-income countries.
Connectivity before AI
The report does not argue that smartphones alone will solve the divide. It also points to digital skills, literacy, online safety, security concerns and the availability of relevant content and services. These issues are familiar across Southeast Asia, where internet access does not always translate into meaningful usage.
But handset affordability is the first gate. Without the device, the rest of the digital economy remains theoretical.
Also Read: The unspoken crisis: Are we building a new digital divide in agriculture?
That makes the GSMA’s report a useful corrective to the current AI conversation. Much of the debate focuses on model capabilities, data centres, regulation and enterprise adoption. Those issues matter. Yet for billions of people, the defining question is more basic: can they afford the phone required to participate?
If governments and the tech industry want AI to be inclusive, they may need to start not with algorithms, but with the low-cost smartphone supply chain. Otherwise, the AI revolution could arrive in emerging markets as another service built for people already connected — while those outside the mobile internet remain exactly where they are.
The post GSMA says smartphone costs could deepen the coming AI divide appeared first on e27.