A Kenyan farmer never opened a bank account, and never needed to. A Shenzhen tech worker isn't scared of AI the way a Seattle one is, and the reason has nothing to do with how much either of them understands about large language models. A sovereign wealth fund in Abu Dhabi just bought enough desert to build a data center campus nine times the size of Monaco. None of these three agreed on anything, yet all three said yes to AI, in three different languages.
That is how a general-purpose technology actually spreads. It does not arrive on one global timeline. It arrives through whatever door is already open in each place, and the door tells you more about what happens next than the technology itself does.
The Phone Already Replaced the Bank in Nairobi
M-Pesa started in 2007 as a way to text a microloan repayment. Nobody designing it was thinking about artificial intelligence. Nineteen years later it serves roughly 40 million monthly users in Kenya alone, more people than have ever held a formal bank account in that country. Mobile money agents now outnumber bank branches by more than 10 to 1 across much of the continent, and by higher ratios in key markets.
WhatsApp rode in on the same rails. Penetration among internet users in Kenya and South Africa now tops 95%. Worldwide, people send more than 7 billion voice notes on the app every day, the format that carries much of how Africa actually uses it, because typing was never the point. Bangladesh went from under 20% smartphone penetration in 2015 to more than 70% today without most of the population ever sitting down at a desktop computer.
So when AI shows up here, it is not a new habit competing for attention against ten other apps. It is the next feature inside the one app that already handles a farmer's loan, a mother's grocery money, and a teenager's first job search. The infrastructure question was solved years ago. AI just has to walk through a door that is already open.
China Isn't Less Worried About AI. It's Worried About Something Else
Stanford's 2026 AI Index put a number on a gap that shows up in every conversation about this topic: 84% of people in China say they're excited about AI, the highest of any country surveyed. In the US, it is 38%. Trust runs the same direction, 72% versus 32%.
Read that as China seeing less risk and you would be wrong. There is a Chinese word for what is actually driving the enthusiasm: involution, the treadmill of running harder just to hold your place, the same word people use for the electric vehicle price war and the food delivery wars. Tiger moms are rushing kids into AI classes because junior robotics and Math Olympiad prep no longer look like enough. After the DeepSeek moment, budget prompting courses flooded Taobao within days. The fear driving Chinese AI adoption is not fear of AI. It is fear of being the one left standing still while everyone else moves.
In the US, the fear points the other way, at the platforms themselves. A decade of social media harming teenagers, surveillance revelations, and a handful of companies holding enormous wealth next to a government that increasingly looks like it shares their boardroom left Americans wanting guardrails, not velocity. Brookings fellow Kyle Chan calls the resulting Chinese strategy “AI-pilled, but not AGI-pilled,” chasing AI inside factories and hospitals rather than the race toward artificial general intelligence that dominates the American conversation. Same technology. Opposite instinct. Both are rational, given what each population actually lived through the last time a wave of technology showed up promising to change everything.
The Gulf Skipped the Consumer Question Entirely
Nobody in Abu Dhabi is trying to win over a skeptical public or ride existing payment rails. The strategy is to own the physical layer outright. G42's Stargate UAE, built with OpenAI, Oracle, Nvidia, Cisco, and SoftBank, is going live with its first 200 megawatts this year, inside a campus designed to eventually hold 5 gigawatts. The full site spans 19.2 square kilometres, close to nine times the size of Monaco, the largest AI infrastructure investment outside the United States. Saudi Arabia's Public Investment Fund-backed HUMAIN is chasing roughly 6 gigawatts of its own by 2034.
Israel is running the same play at higher density. Nvidia is building a major research campus in Kiryat Tivon, designed to house thousands of engineers.
Israel's government treated data centers as strategic infrastructure in February 2026 and doubled its power plant construction targets to keep pace. This is not adoption in any consumer sense. It is a bet that owning the compute, not the app on top of it, is what buys leverage over the next decade: export capacity, sovereignty, a seat at the table when the rules get written somewhere else.
The Door Tells You the Next Move
None of these three sits ahead of or behind the others on a shared curve. A bank in Nairobi, a policy debate in Washington, and a sovereign fund in Riyadh are not variations on the same story. They are three different stories that happen to be using the same underlying technology.
Build a rollout plan for one and drag it unchanged into another, and it breaks on contact.
A product built for American-style skepticism will undersell itself in a market where enthusiasm already outruns the product. A rollout built on desktop-era assumptions will completely miss a market where the phone is already the bank, the ID card, and the storefront. Know which door you are walking through before you decide what to bring.
Before your next regional AI rollout, name the door your users are already walking through in that market — a messaging app, a trust deficit, a sovereignty mandate — and build the entry point to match it, rather than exporting the plan that worked at headquarters.
Sources:
Infobip. "WhatsApp Statistics 2026: Global Usage & Market Overview." infobip.com, 2026.
GSM Association. "State of the Industry Report on Mobile Money." gsma.com, 2024/2026 updates.
Safaricom / Kenyan reporting on M-Pesa monthly active users, 2026.
Shao, Grace. "Why Chinese Citizens Are Far More Optimistic About AI Than Americans." Bloomberg, 14 Aug. 2026.
Stanford HAI. AI Index Report 2026.
Chan, Kyle. Brookings Institution commentary and related interviews, 2026.
G42. Stargate UAE announcements and campus specifications, 2025–2026.
Nvidia / Calcalist / Times of Israel reporting on Kiryat Tivon campus, 2025–2026.
Israeli government announcements on data center acceleration, February 2026.
