The AI Divide Is No Longer About Access to the Internet | ✉️ #98
Hey! 👋
For years, the digital divide was understood as a problem of infrastructure. Some countries had fiber, reliable electricity, modern computers and fast mobile networks; others did not. But AI is creating a different kind of divide. In countries like Venezuela, the problem is no longer simply that the technology arrives late. In some cases, it is actively unavailable because major companies such as OpenAI and Anthropic do not officially provide access to their services there.
This matters because AI is quickly becoming more than a consumer product. It is becoming part of the basic infrastructure of software development, education, research and business. A developer in Madrid or New York can open an account, obtain an API key and start building with the latest models within minutes. A developer in Caracas may have the same laptop, the same skills, the same internet connection and the same ability to pay, but still be excluded because of the country from which they are connecting.
The consequence is a form of technological inequality that is much harder to solve locally. Venezuela can improve its fiber networks, universities can train better engineers, and companies can invest in modern hardware, but none of those things can change a provider's list of supported countries. The barrier is not physical infrastructure. It is a commercial, regulatory or risk decision made somewhere else.
This also creates a compounding disadvantage. AI does not just help people complete tasks faster; it helps them learn faster, write more software, test more ideas and build more ambitious products. If developers and companies in one country spend five years working with the most advanced AI tools while those in another country face restrictions, the difference at the end of those five years will not simply be five years of missed access. It will be five years of accumulated productivity, knowledge and experimentation.
Of course, people find ways around these restrictions: foreign companies, foreign payment methods, VPNs, intermediaries and accounts registered abroad. But that creates another tax on innovation. Instead of spending time building products, entrepreneurs spend time building structures that allow them to access the same tools their competitors can use directly. It adds cost, legal uncertainty and operational risk to companies that are already competing from weaker economic environments.
The next digital divide may therefore have very little to do with who has internet access. It may be about who is allowed to access the intelligence running on top of it. AI has already reached countries like Venezuela in the sense that people understand it, study it and want to build with it. The problem is that parts of the AI economy remain behind geographic gates. And a technological gap created by infrastructure can eventually be solved by building infrastructure; a gap created by someone else's permission is much harder to close.
What We've Discovered
Scaling Terraform Across many Teams: A Native Framework for Platform Engineering: Very nice approach to split the buildings blocks created by platform team, from configuring them as a platform user - and it doesnt' require much additional tooling.
Are Repository-Level Context Files Helpful for Coding Agents? "Overall, our results suggest that context files have only marginal effect on agent behavior, and are likely only desirable when manually written."
How Generative and Agentic AI Shift Concern from Technical Debt to Cognitive Debt: Generating vast amounts of new code and systems creates a threat bigger than technical debt: the team might lose any understanding of what the system is meant to do and why it was designed them way, thus losing the ability to further improve or change it.
The 99th mkdev dispatch will arrive on Friday, September 11th. See you next time!