Silicon Valley Splits Over Closing the Borders to Chinese A.I

By Maya Ellison · Reporting from Detroit · July 26, 2026

The whole debate over Chinese AI models and "distillation" is a spectacular piece of theater designed to distract us from one thing: concentrated power.

When 'National Security' Becomes Market Control

The whole debate over Chinese AI models and "distillation" is a spectacular piece of theater designed to distract us from one thing: concentrated power. We are being told this week that our access to advanced technology—the very engine of progress—is somehow jeopardizing national security, all because Moonshot AI’s Kimi K3 has made open-weight models cheaper and more capable than the proprietary systems built by Anthropic or OpenAI.

The facts are undeniable: Distillation is a legitimate technique, as confirmed by Shashi Bellamkonda. It's how smaller, efficient models learn from larger ones. But what we’re seeing isn't a national security crisis; it’s an oligopoly attempting to regulate its own playground. Anthropic and OpenAI are aggressively lobbying regulators (vogelitlawblog.com) because their business model relies on selling access to closed, proprietary systems. They claim Chinese labs illegally harvested data—a process Michael Kratsios noted with the allegation that Moonshot AI distilled Anthropic’s Fable.

The Open Source Argument vs. Corporate Panic

The corporate panic is palpable. On one hand, you have Nvidia and Microsoft executives (vogelitlawblog.com) arguing that open-source models must remain free for everyone to build upon—a stance echoed by the thousands of startups who signed a letter opposing outright bans on Chinese AI models (cryptobriefing.com). On the other hand, you have the major players demanding restrictions, citing IP theft and national security risks.

The truth is that concentrated wealth buys concentrated power. When tech giants scream about "unfair advantage" or "illicit distillation," they are really screaming because their market share is threatened by competition that costs a fraction of the price (cryptobriefing.com). The fact remains: Chinese models have captured over 30% of US developer token usage since February 2026, and they do it at $2 to $3 per million tokens—a staggering blow to the cost-of-entry barrier that U.S. corporate AI has maintained for years.

No verified historical parallel survived fact-checking for this story.

Commoditizing the Frontier Model

The argument that open models are dangerous is a scare tactic designed to keep us buying expensive, locked-down subscriptions. The reality—as Alvin W. Graylin noted when Kimi K3 "commoditized frontier models"—is that innovation thrives on shared knowledge and competition. When you restrict access to powerful, affordable tools, you aren't protecting the public; you are protecting venture capital investments.

This whole debacle is not about who stole what data, but about who controls the pipes of future economic growth. The lobbying efforts by closed-source AI companies must be resisted because they seek to erect new digital paywalls that will inevitably raise costs for developers and consumers alike, making advanced technology a luxury good rather than an infrastructure right.

Silicon Valley Splits Over Closing the Borders to Chinese A.I

When 'National Security' Becomes Market Control

The Open Source Argument vs. Corporate Panic

Commoditizing the Frontier Model

Sources