None of this should feel unfamiliar if you lived through telecom, mobile, or cloud.
In the early 2000s, U.S. carriers convinced themselves that owning the network meant owning the future. They optimized for ARPU, control, and contractual lock-in. Meanwhile, VoIP app companies, OTT operators, and Skype all focused on deployment speed, adoption depth, and scale economics. When VoIP rode over the top, the advantage flipped almost overnight.
Mobile followed the same script. Closed platforms promised quality and safety. (Do you remember Symbian? Blackberry??) Open platforms promised reach. Android did not win because it was elegant. It won because it spread faster than control systems could contain it. Apple won by making things simple. For the iPhone, it was the interface that spurred an ecosystem (The App Store). By the time incumbents reacted, the ecosystem had already consolidated around scale.
Cloud repeated the lesson again. Enterprises clung to bespoke infrastructure and private stacks long after hyperscalers had crossed the reliability threshold. Cost curves did the rest. Control delayed adoption. It did not prevent it.
AI is now replaying those cycles in compressed time.
Closed, premium models look strong early. They dominate demos, headlines, and margin narratives. Open systems look rougher. Less polished. More chaotic. Then they spread. They attract developers. They collapse costs. And suddenly, the question is no longer performance at the top end. It is who can deploy everywhere.
That is the inflection point China is pressing into, why, I’ve been featuring news from the South China Morning Post in my almost daily newsletter, The Comunicano.
Distribution Beats Prestige Every Time
Telecom history is brutal on this point.
Standards that spread beat standards that negotiate. Platforms that tolerate variation beat platforms that enforce purity. Ecosystems that accept messiness beat ecosystems that demand permission.
Chinese AI labs are not chasing prestige positioning. They are chasing ubiquity. Open weights. Local deployment. Hardware adjacency. Device integration. That is not academic ambition. It is a platform strategy.
The same forces that crushed circuit-switched voice, proprietary mobile OS’, and closed cloud stacks are now applying pressure to AI. Not gradually. All at once.
Control Is a Temporary Advantage
Every incumbent believes control is a moat until it becomes friction.
Telecom operators learned this when over-the-top players bypassed billing systems. Platform owners learned it when developers routed around app store constraints. And, the Cloud vendors learned it when abstraction layers erased hardware differentiation.
AI incumbents are now learning the same lesson. Control buys time. It does not buy inevitability.
China’s approach accepts lower margins early in exchange for a strategic position. That trade looks reckless if you model quarterly revenue alone. It looks inevitable if you have studied platform history.
The Pattern Is the Point
This is not about nationalism. It is not about ideology. It is not even about AI per se.
It is about how complex technology markets consolidate.
Open distribution wins when compute becomes abundant. Integration wins when hardware and software converge. Infrastructure wins when demand outpaces optimization. Talent migrates when the work is real.
Telecom told us this. Mobile confirmed it. Cloud locked it in.
AI is simply the next layer.