The AI boom was supposed to democratize technology.
Instead, one of its unintended consequences is becoming increasingly difficult to ignore:
AI is making basic technology more expensive for the people who can least afford it.
The world's cheapest smartphones are disappearing.
Not because consumers suddenly stopped wanting them.
Not because smartphones became technologically obsolete.
And not because manufacturers forgot how to build cheap devices.
The problem is more fundamental.
The memory chips that make smartphones possible are increasingly being redirected toward artificial-intelligence data centers.
And the economics are brutal.
Counterpoint says global smartphone retail prices have risen about 15% so far in 2026, while newly launched models are roughly 25% more expensive than their year-earlier counterparts. In price-sensitive markets, the increases are even worse: India around 21%, Asia-Pacific 19%, and the Middle East and Africa 18%.
This isn't simply an expensive-phone story.
It is becoming a digital-access story.
The $100 phone is disappearing
For years, the smartphone industry followed a powerful formula:
Make hardware cheaper.
Sell millions of units.
Get billions of people online.
Chinese manufacturers became particularly good at this.
Xiaomi, OPPO, vivo, Huawei and other manufacturers helped turn smartphones from luxury products into mass-market computing devices.
In 2025, more than one in four smartphones shipped globally cost less than $150.
But the bottom of the market is now collapsing.
IDC says 173 million smartphones priced below $100 were shipped in 2025.
In the second quarter of 2026, shipments in that category fell by almost 60% year over year. IDC says manufacturers are cutting low-end models and shifting toward higher-margin products as memory costs explode.
Counterpoint's analysis points in the same direction: memory costs for low- and mid-range smartphones have become an enormous share of the bill of materials, forcing manufacturers to raise prices and reconsider the economics of cheap devices.
This is not merely a temporary shortage.
It is becoming a question of what manufacturers choose to build.
And increasingly, the answer is:
fewer cheap phones.
Why does AI need so much memory?
This is where the story gets interesting.
When people hear “AI chips,” they often think about Nvidia GPUs.
But AI data centers require enormous amounts of memory too.
Modern AI accelerators use high-bandwidth memory, or HBM, to move huge volumes of data rapidly between processors and memory.
The same broad semiconductor industry is also responsible for producing memory products used throughout consumer electronics.
Samsung, SK Hynix and Micron dominate global memory production, accounting for more than 90% of the market according to industry estimates.
The economics are obvious.
If a manufacturer can sell high-value memory into an AI data center rather than lower-margin consumer electronics, the AI customer becomes extremely attractive.
IDC says memory costs have risen more than 300% year over year, while low-end smartphones are being hit hardest.
That creates a strange technological paradox:
The world is investing enormous amounts of money to make AI cheaper and more accessible while simultaneously making one of the primary devices needed to access that AI more expensive.
The AI divide could become the next digital divide
This is where Africa becomes particularly vulnerable.
The global conversation about AI often starts with:
Who has the best models?
Who has the biggest data centers?
Who has the fastest GPUs?
Who has the most powerful agents?
But there is another question:
Who can afford the device required to use any of it?
The GSMA's 2026 State of Mobile Internet Connectivity report is blunt.
More than 3.4 billion people do not use mobile internet, even though more than 90% of them live within mobile broadband coverage.
That means the problem increasingly isn't whether a network exists.
It is whether people can afford to participate.
The GSMA says handset affordability is the single biggest barrier to mobile internet adoption in surveyed low- and middle-income countries.
For the poorest 20% of people globally, an entry-level internet-enabled phone cost the equivalent of 44% of average monthly income at the end of 2025.
In sub-Saharan Africa, the figure reached 76%.
Think about what that means.
A device marketed as an “entry-level” smartphone can effectively represent most of a poor person's monthly income.
And now the component costs are rising.
Africa is especially exposed
Africa doesn't have a huge premium-smartphone cushion.
The continent's smartphone market is overwhelmingly concentrated at the affordable end.
Omdia says 81% of smartphones shipped in Africa in 2025 cost below $200.
That means the continent's smartphone ecosystem depends disproportionately on the very market segment being squeezed by rising component costs.
And Nigeria is particularly important.
Nigeria is one of the markets identified by the GSMA for a 2026 pilot programme aimed at introducing affordable 4G smartphones.
The other initial markets include the Democratic Republic of Congo, Ethiopia, Rwanda, Tanzania and Uganda.
The irony is striking.
The industry is trying to make a $40 smartphone possible at exactly the moment when the economics of ultra-cheap smartphones are becoming more difficult.
Nigeria doesn't need a more expensive digital divide
Nigeria already has a connectivity paradox.
Millions of Nigerians live within mobile broadband coverage.
But being covered by a network is not the same thing as being connected to the digital economy.
A person without an affordable smartphone cannot fully participate in:
- online banking;
- digital payments;
- e-commerce;
- remote work;
- online education;
- telemedicine;
- government services;
- social media;
- digital entrepreneurship;
- AI tools.
And increasingly, access to AI itself.
This matters because AI is becoming embedded in ordinary economic activity.
Businesses are using AI for customer service.
Students are using AI for education.
Developers are using AI for programming.
Freelancers are using AI for productivity.
Small businesses are using AI for marketing.
Governments are beginning to digitize services.
If the entry price for participating in that economy rises, the people at the bottom don't simply buy a slightly more expensive phone.
Some never enter the economy at all.
This is how technology inequality compounds
Imagine two young Nigerians.
One can afford a ₦500,000 smartphone.
Another can only afford a ₦100,000 device.
The first gets access to modern apps, AI assistants, cloud services, digital banking and increasingly sophisticated productivity tools.
The second may be forced into an older, slower or lower-storage device.
Then the device fails.
Replacement costs rise.
The user delays buying another.
Eventually, they go offline.
The difference isn't simply between two phones.
It becomes a difference in:
education → productivity → income → opportunity.
That is how a hardware price increase becomes an economic inequality problem.
And AI could make the problem worse
The AI industry is entering an extraordinary capital-spending cycle.
S&P Global estimates U.S. hyperscalers could spend more than $1.3 trillion in capital expenditure in 2027, compared with about $870 billion projected for 2026 and $470 billion in 2025.
That means enormous demand for:
GPUs.
HBM.
DRAM.
NAND.
Networking equipment.
Power.
Cooling.
Data-center construction.
Land.
Electricity.
The AI industry isn't simply competing with smartphones for attention.
It is competing for industrial capacity.
And the companies selling into AI infrastructure often have considerably more purchasing power than companies trying to build a $70 smartphone for a low-income consumer.
This is capitalism doing exactly what capitalism does:
scarce resources move toward the highest-paying customer.
The consequence is that the world's poorest consumers can lose access without anyone explicitly deciding to exclude them.
Samsung's profits tell the story
The semiconductor industry's current economics make the imbalance even clearer.
Samsung recently projected approximately $80 billion in quarterly operating profit, driven largely by the AI-fueled memory boom.
Memory demand is producing extraordinary profits for chipmakers.
At the same time, Samsung's mobile division reported a much weaker performance.
This is not a contradiction.
It is the point.
The same company can benefit enormously from AI infrastructure while consumer electronics face rising component costs.
The AI boom is creating winners inside the hardware supply chain.
But not every hardware customer is a winner.
The cheap phone was one of the greatest digital-equality technologies
This part of the story is often overlooked.
The smartphone didn't democratize computing simply because it was powerful.
It democratized computing because its price kept falling.
A person who could never afford a laptop could buy a smartphone.
A street trader could receive payments.
A student could watch lectures.
A small business could advertise on Facebook.
A freelancer could work with clients overseas.
A farmer could access information.
A young person could create a YouTube channel.
The smartphone became the computer of the Global South.
That is why the disappearance of the cheapest devices matters so much.
We are not talking about whether somebody gets the latest camera.
We're talking about whether somebody gets a computer at all.
The African AI dream has a hardware problem
Africa is being told that AI could transform education, healthcare, agriculture, finance and government.
That may be true.
But there is an uncomfortable prerequisite:
People need affordable ways to access it.
You cannot close an AI divide while allowing the hardware divide to widen.
You cannot tell a Nigerian student:
“Learn to use AI.”
while the student's family cannot afford a device capable of running modern applications comfortably.
You cannot tell a small African business:
“Become AI-powered.”
while the owner's smartphone is running out of storage and struggling with basic applications.
And you cannot build an inclusive digital economy if the price of entry keeps moving upward.
The answer isn't necessarily cheaper new phones
There is another opportunity here.
If new smartphones become structurally more expensive, governments, telecom companies and manufacturers need to think beyond selling new devices.
Refurbishment could become strategic infrastructure.
A good refurbished smartphone can extend the useful life of hardware.
Trade-in programmes can move used devices down the income ladder.
Repair ecosystems can keep phones alive longer.
Device financing can reduce the upfront cost.
Local assembly can help in some markets, although assembly alone does not solve the global memory-price problem.
And governments could support targeted affordability programmes instead of simply subsidising expensive flagship devices.
Africa needs a functioning secondary smartphone economy.
Because if the new-phone market moves upward, millions of people will depend on the devices that already exist.
There is also a sovereignty question
Africa consumes enormous amounts of imported technology.
That means a decision made in a semiconductor factory in South Korea, Taiwan, the United States or elsewhere can affect whether a Nigerian student can afford to participate in the digital economy.
This is the deeper lesson.
Digital sovereignty isn't only about where your data is stored.
It is also about whether you have affordable physical access to the digital world.
If Nigeria builds sophisticated AI policies but its citizens cannot afford the devices required to use AI, the policy is incomplete.
If Africa wants an AI economy, it needs an affordable hardware economy underneath it.
The market is already responding
Consumers are delaying upgrades.
Manufacturers are reducing storage configurations.
Some are reducing camera specifications.
Some are pushing more 4G models.
Others are moving toward higher-margin devices.
Counterpoint says more than 40% of smartphone models have seen price increases in 2026.
IDC expects the global smartphone market to contract sharply this year, with low-end Android devices among the biggest casualties.
That is a remarkable market transformation.
The smartphone industry spent more than a decade making computing cheaper.
Now it is learning how to sell fewer, more expensive devices.
The uncomfortable question for the AI industry
AI companies love to talk about democratization.
AI will supposedly make knowledge cheaper.
AI will increase productivity.
AI will give everyone access to expert-level capabilities.
But there is a contradiction hiding underneath that narrative.
If AI infrastructure consumes scarce components, drives up memory prices and contributes to the disappearance of affordable smartphones, then the benefits of AI may be distributed very unevenly.
The richest companies get enormous computing clusters.
The richest consumers get premium AI devices.
The poorest consumers lose the cheapest devices.
That isn't automatically the AI industry's fault.
The semiconductor industry has its own supply constraints, investment cycles and capacity decisions.
But it is the predictable consequence of putting extraordinary demand into a supply chain that cannot instantly expand.
And policymakers need to recognize it.
The $20 and $30 smartphone dream is now under pressure
The GSMA had identified extremely low-cost smartphones as a potentially transformative solution.
Its analysis found that reducing entry-level smartphone prices to around $30 could make devices affordable to roughly 1.6 billion additional people, while a $20 price could potentially reach around 2.2 billion people living within mobile broadband coverage.
That was the opportunity.
The memory crisis threatens to move the market in the opposite direction.
Instead of:
$100 → $70 → $50 → $30
we could see:
$100 → $130 → $150 → $200.
And once that happens, the people who were supposed to be the biggest beneficiaries of the digital revolution become the people least able to afford entry.
AI should not become a luxury layer on top of a digital exclusion crisis
There is nothing wrong with building powerful AI infrastructure.
The world needs it.
AI can generate enormous productivity gains.
But technological progress should not be measured only by how powerful the machines in a data center become.
It should also be measured by how many ordinary people can afford to participate in the economy those machines are creating.
Africa needs to ask a harder question than:
“How do we become an AI leader?”
The first question should be:
“How do we make sure Africans can afford to enter the AI economy?”
Because if the cheapest smartphone disappears, the problem isn't merely that someone's next phone costs more.
For millions of people, the problem is that there may be no next phone at all.
And that could turn the AI revolution into something deeply paradoxical:
The technology promising to democratize intelligence may simultaneously make access to the digital world less democratic.
Comments
Post a Comment