For the fourth time this year, an engineer is rewriting the same compliance script somewhere in a data center outside of Frankfurt. Brussels has new regulations. Delhi has new regulations. A new Beijing memo that no one has yet to fully comprehend. It used to be easier. Once built, it can be shipped anywhere. Quietly, that model is dying.
Splinternet is the word that people keep reaching for. It sounds like jargon, and for years it was only discussed by think tanks and a few anxious Google executives. However, the fracturing is no longer hypothetical. The most obvious example is China’s Great Firewall, which is cited by all. The less evident examples are those that are changing the economic landscape, such as Russia’s division of RuNet, India’s insistence that payment data remain on Indian territory, and Kazakhstan, Vietnam, and Indonesia all creating their own versions of the same regulations. Even the European Union, which is by no means an authoritarian organization, has erected a regulatory barrier so strong that American cloud service providers now create completely different stacks specifically for Hamburg clients.
All of this has a real cost, but it’s the type of cost that is hidden by internal line items. According to an earlier estimate by the Information Technology and Innovation Foundation, data-localization regulations could cost the US cloud computing sector alone between $22 billion and $35 billion in lost revenue. That figure is now nearly ten years old. The more recent estimates are more expensive and do not account for the more difficult-to-measure factors, such as the engineering hours, redundant infrastructure, and deals that are never completed because the legal review process takes nine months. Global flows of investment, goods, and services could triple by 2025, according to a McKinsey forecast. A portion of that growth was visible. Many of them didn’t. How much friction the splintering added is difficult to ignore.
Speaking with those who manage large-scale platforms gives me the impression that the days of having a single product for the entire world are over. These days, a social network is any version that can be salvaged for the American market, including Turkish, Brazilian, and EU versions with their own moderation guidelines. Every version requires a separate legal team, content reviewers, and domestic servers. The economies of scale that allowed businesses like Microsoft, Amazon, and Meta to become so powerful are still present, but they are gradually being undermined from all sides.
A portion of this is actually defensive. When European leaders began discussing regional clouds following Edward Snowden’s revelations regarding NSA surveillance, they weren’t acting paranoid. Many observers at the time considered Angela Merkel’s demand for a European internet—which was hidden from American view—to be alarmist. It no longer sounds alarmist. In order to avoid US infrastructure, Brazil installed an underwater cable. Germany persisted in its efforts. Around 2013, the trust that supported the open web—the belief that data traveling across borders was secure—broke and was never fully restored.
What is replacing it appears to be a patchwork of intranets masquerading as an internet. In Karachi or Lagos, you can still use a browser to access the majority of the world. However, how and what you arrive at depends more and more on which jurisdiction’s attorneys arrived first. Tim Berners-Lee’s dream of a web that was owned by no one and belonged to everyone was never going to come true. America’s strength and lack of interest allowed it to succeed. Both of those circumstances have become less severe.
Whether the splintering will continue is no longer a question. Yes, it does. What is lost along the way and if anyone is keeping track are the questions. It appears that investors think the giants will change. Perhaps they will. However, adaptation is costly, and eventually someone has to foot the bill.
