Trust Wars : Kimi and the Cognitive Domain War

China went open. America went closed. The AI race just became a battle for your trust 

LAOWAI - Tyler Johnson 

 

 

Huawei makes better phones than most people give it credit for. Lenovo has been the world’s biggest PC maker for over a decade. And yet neither owns a room the way Apple or Samsung does. There is a ceiling that China’s tech giants keep hitting, and it’s not a tech problem……it’s a trust problem.

China’s playbook has worked for 20 years…….protect the home market…..subsidize the champions and let 1.4B consumers and local governments do the heavy lifting…..then push outward. It built Huawei into the largest telecom equipment maker on earth. And then Huawei hit the wall. The Five Eyes froze it out of their 5G cores. Germany is finishing the job this year, ripping Huawei and ZTE gear out of core networks by the end of 2026. A decade of denials hasn’t mattered because the company sits under a legal system that can compel cooperation with state intelligence, behind an ownership structure nobody outside China can verify. Lenovo’s has the largest market share for PCs in the world, solely driven by domestic demand. 

Cisco and Ericsson build the same gear at higher prices and have never faced a comparable wave of bans(except in China!). Trust is a currency, and America is spending its own down too. European governments are quietly migrating hundreds of thousands of workstations off Microsoft and moving data out of U.S. clouds, and the reason should sound familiar……the CLOUD Act lets Washington reach data held by American companies anywhere on earth. Berlin is starting to price American compellability the way it priced Chinese compellability. The difference is the structure. U.S. companies can be sued, audited, and dragged in front of Congress in public. It was never about who builds the better product. It's about who the company answers to, and whether an outsider can ever really check.

Moonshot AI dropped Kimi K3 this week. The largest open-weight model ever released at 2.8 trillion parameters. It’s also priced at a fraction of the U.S. frontier labs. The largest open weight model for US firms is ~400B (Meta). Open weights are the billions of numbers the AI learned during training, the closest thing it has to a brain. Publish them, and anyone on earth can download that brain, run it on their own machines, modify it, and see for themselves what it does. Keep them locked up, and all you can do is send a question into someone else’s server and trust the answer that comes back. That’s what makes Kimi different from every Chinese tech expansion before it. You couldn’t inspect a 5G core. You can inspect a weights file.

The irony….China built its rise on closed systems….a closed internet, closed politics, and protected champions. Now, in the one domain that might finally get it past the trust ceiling, China is playing open. Meanwhile America, the open-markets country, has locked its frontier AI in black boxes. OpenAI, Anthropic, and Google don’t publish their best weights; you send a prompt in and trust what comes back. Chinese labs now account for the majority of open-weight downloads worldwide. China is out-opening America at America’s own game.

You could argue this is a data grab. Flood the world with free models and harvest everything. But it doesn’t hold up because a self-hosted model never has to phone home to China at all(or at least until the call comes). The real play is slower and smarter. Get embedded in enough startups, universities, and governments, and you become the default. The default becomes the standard. And the standard decides how a few billion people’s AI answers questions……including the ones Beijing would rather not answer. It’s soft power, the same kind Hollywood spent a century exporting. We should recognize the move. We invented it.

NATO has a name for this terrain. Its Allied Command Transformation has spent years studying cognition as a sixth domain of warfare (after land, sea, air, space, and cyber). Their language is blunt: trust is the targeted vulnerability, and cognitive warfare works not by force but by shaping the tools people think with. If that’s right, the trust ceiling was never a branding problem. It was the front line of a war that’s been running for years, and AI just made it visible. Every model a billion people route their thinking through is a foothold. Every cheap open-weight release is also an offer: trust us instead.

So what does America do? First, drop the panic. Competition is the one thing open-market systems are actually built for……iron sharpens iron, and a serious rival is exactly what forced DeepSeek-level efficiency out of American labs’ complacency on pricing. The threat isn’t that China is competing. It’s that America is refusing to compete on the open field where the adoption war is being decided. Export controls buy time, not victory, and every month behind the wall is a month lost. The fix is structural. Tie federal research money and government AI contracts to open licensing: publish the weights or don’t take the check. Make open release commercially viable for the American labs that aren’t OpenAI. Right now Meta is nearly alone, a one-company ecosystem against China’s five. And narrow the export regime to actual military end uses so American compute flows into the global open ecosystem instead of ceding it by default. None of this needs new bureaucracy. It needs Washington to notice that openness is a strategic asset America is letting China own for free.

I spent ten years in China as a locally hired foreign national inside a multinational, building the infrastructure that let a Western company operate on the ground. That seat gave me a side-by-side view……same market, two operating systems. The multinational answered to auditors, shareholders, courts that could actually be sued. Every one of those checks slows you down….but, this is why you can plug into that system without wondering who’s really running it. Open markets don’t win because they’re virtuous. They win because they force information into the open, and that friction is what lets human capability compound instead of bending around a chain of command nobody outside can see. Kimi K3 is real, and it’s good. But only one side of this race has the structure built to eventually make you show your work.

So look at what you’re running. Ask who owns it, who can compel it, and what happens when you can’t check. That’s the only defense in a war fought in the sixth domain. Nobody can do that looking for you.

God Bless America