Why U.S.--China AI Diplomacy Remains Performative and Ineffective

The U.S.-China AI summit is a high-stakes performance of geopolitical theater, but the real-world impact will likely be near zero. While industry leaders and politicians chase the prestige of a grand deal, the structural reality of mutual distrust and conflicting incentives ensures that any agreement remains purely symbolic. For observers and investors, the advantage lies in ignoring the headlines about safety frameworks and instead tracking the quiet, incremental shifts in communication channels and private-sector competition. The danger is not just that these talks will fail; it is that they create a false sense of security, masking the reality that the race for AI dominance continues, driven by private-sector interests that prioritize speed over the very safety protocols being debated in Washington.

The Illusion of the Grand Deal

The current diplomatic push for an AI safety agreement between the U.S. and China suffers from a basic misunderstanding of how these systems interact. The prevailing narrative that a grand deal can be struck to pace the frontier of AI development ignores the fact that neither nation trusts the other enough to make such an agreement enforceable. As Ian Kreitzberg notes, the last major attempt at a tech-related pact, the 2015 cyber-agreement, fell apart within three years.

> "I did not speak to a single person who thought, yeah. They've got this, they're going to handle it and we've gonna have some sort of deal agreement partnership on AI safety between the US and China."
>
> -- Ian Kreitzberg

The system responds to this lack of trust by defaulting to performative diplomacy. When politicians push for broad safety frameworks, they are addressing the optics of fear rather than the mechanics of risk. The consequence is a cycle of agreements that lack any real enforcement, creating a facade of progress that poisons the well when the inevitable non-compliance occurs.

Where Immediate Pain Creates Lasting Moats

The industry's internal conflict reveals a deeper truth about competition. While CEOs like Sam Altman and Dario Amodei publicly advocate for pacing the frontier, their private incentives are heavily influenced by the rise of Chinese open-weight models. These models, such as those from DeepSeek, are eroding the margins of American firms by providing cost-effective alternatives for enterprise users.

The safety conversation is, in part, a strategic attempt to use regulation as a competitive moat. If U.S. firms can lobby for global safety standards that effectively slow down Chinese development or restrict the proliferation of open-weight models, they protect their market share.

> "Their biggest competition right now is free open weight systems from Chinese developers. If you're Sam that's a concerning thing. If you're Mark that's a concerning thing, right? These are if those proliferate and if those are allowed to proliferate that's gonna hurt your margins."
>
> -- Ian Kreitzberg

This creates a feedback loop: the more the U.S. pushes for regulation, the more the Chinese system, which operates through a unique public-private partnership, doubles down on its own development to maintain its 15-year plan for global dominance. The attempt to solve the problem via regulation actually accelerates the competitive pressure that drives the race.

The Pragmatic Pivot: Incrementalism Over Grandiosity

Because grand deals are doomed, the only meaningful outcomes from this summit will be the most boring ones: incremental communication channels. Moving from high-level rhetoric to a functional, bureaucracy-compatible email system for incident notification is a trivial step, but it is the only one that acknowledges the reality of the Chinese administrative system.

The system logic here is clear: the most durable solutions are those that require the least amount of trust. By identifying overlapping interests, where it is in the independent interest of both the U.S. and China to avoid specific AI-driven catastrophes, the two nations can achieve stability without needing to rely on a handshake that can be discarded by the next administration.

Key Action Items

• Filter the Noise (Immediate): Ignore headlines regarding AI safety breakthroughs from the summit. These are designed for public consumption and lack the structural teeth to alter development trajectories.
• Monitor Communication Infrastructure (Next 3-6 months): Track whether a formal, low-level notification channel, such as the proposed email system, is established. This is the only indicator of a functional, rather than performative, diplomatic outcome.
• Analyze Open-Weight Adoption (6-12 months): Watch the enterprise adoption rates of Chinese open-weight models. If these models continue to gain market share, expect U.S. firms to intensify lobbying efforts for export controls and restrictive safety regulations.
• Focus on Private-Public Dynamics (12-18 months): Observe how Chinese companies like DeepSeek and Alibaba evolve. Their ability to integrate state resources with private agility is the true variable in the AI race, not the public posturing of their leaders.
• Assess Policy Durability (Ongoing): Recognize that agreements made by the current administration are subject to rapid reversal. Invest in strategies that assume a volatile regulatory environment rather than one defined by stable, long-term international treaties.