Jennifer Valenti, Co-founder & Chief Strategy Officer, Cited
February 4, 2026
I've been thinking about this for over a decade now. Not exactly assembled in the way Phil Fersht lays out so well in his recent analysis, but directionally (how we think about U.S. sovereignty, our perceived advantages in technology, and particularly in AI).
Phil's piece is important. He's right that the U.S. is betting everything on AGI while China builds the infrastructure layer that will power global AI adoption. But I want to push his analysis further. While Phil frames this as a U.S.-versus-China competition where infrastructure beats innovation, I don't believe it's truly an us-versus-them scenario any longer. And understanding why this matters enormously for anyone thinking about how content gets discovered in an AI-driven world.
The Chokepoint Era Is Over
Post WWII the world changed. Through the strategic genius of FDR, Truman, and Eisenhower, the US branched across the world through a system of treaties, alliances, and organizations that cemented US dominance through the enablement of permanent global military presence. These international bases became chokepoints across every dimension of warfare - land, air, and sea.
Before WWII, we were sovereign and naval, only. Now, post-Pearl Harbor, it became very clear that we needed to metastasize our military might across every sea and continent possible. Through the installation of foreign US bases, we cemented ourselves as the dominant, global force whose mere presence afforded a power not yet seen. This multi-decade strategy, forged from the best military minds in modern history, laid a legacy of extraordinary control and led to seemingly permanent and ubiquitous US dominance. Control that would last for nearly a century. Until now.
AI is leveling the playing field. It's removing the veneer of control over perceived chokepoints and untethering the world from the U.S.'s historical control. Phil captures this brilliantly when he describes how China's open-weight models are becoming the default infrastructure layer while American frontier models remain "expensive luxury products accessible primarily to wealthy markets."
And yes, we have overplayed our hand, exactly as Phil articulates. But I'm not yet ready to lift the arm of China in victory.
What the U.S.-vs-China Frame Misses
Here's what so many are still underestimating: it's not about which country's infrastructure wins. It's about what happens when users get their hands on all of it.
Consider the most extraordinary example we have seen yet: the launch of OpenClaw (formerly Moltbot, formerly Clawdbot, formerly “I’m working on something big, I’ll eat dinner later”.) This one, seemingly simple upgrade from the LLMs we are now used to has quite literally changed everything. All of a sudden, we took the agent out of the box and gave it access to... everything. Browsers, messaging services, private servers, calendars, inboxes, social media accounts.. whatever you were capable of programming it to point to, it could access.
This didn’t happen in a year. It happened... overnight. In just one week there were millions of lobster agents running amok across mac minis and linux servers everywhere. And, any time an agent encountered friction, a human solved it. In the time it took Yiheng Wang to solve the Rubik’s Cube, humans pushed through every sovereign technical barrier that existed.
It took all of these models working together to increase the power of these agents while decreasing compute power, token pulls, and system limitations. It wasn't China that did that. It wasn't the U.S. It was the world of users across all platforms, brilliantly challenged with overcoming limitations and bridging and patching and sewing together the world's network of AI. Working together through orchestrated harmony built through creativity, will, and determination, the lowly individual showed undisputable might. American, Indian, Chinese, Russian, Kenyan, French, Irish, Saudi, Australian users worked together. Worked independently. And it worked.
Chokepoints vanished. Sovereign advantage vanished. All that remained was the unique human ability to create.
The Multi-Model Reality Changes Everything
Phil is right that we can't optimize for a single platform anymore. But the deeper implication is that the future of AI search isn't about which infrastructure wins. It's about orchestration across all of them.
Users are already doing this. They're not loyally asking ChatGPT a question and accepting the answer. They're asking ChatGPT, then Claude, then running the same query through a local model. They're using one AI to fact-check another. They're chaining outputs together, combining the reasoning capabilities of frontier models with the efficiency and privacy of open-weight alternatives.
This is the multi-model reality. Your content might be accessed by GPT-4 in San Francisco, DeepSeek running locally in Mumbai, Qwen customized for regional languages in Jakarta, or all of the above in a single orchestrated workflow by a developer in Nairobi.
Why Authority Becomes the Only Strategy
At Cited, we've been focused on this exact challenge: how do you build content that remains discoverable, credible, and valuable when the infrastructure layer fragments across competing platforms, geographies, and model architectures?
The answer isn't optimizing for Chinese models or American models. The answer is genuine authority.
When users orchestrate across multiple models, when AIs cross-reference each other's outputs, when the search landscape becomes a network instead of a hierarchy, the content that consistently surfaces is the content that deserves to. Not because it gamed one algorithm, but because it's genuinely valuable enough that every system recognizes it as authoritative.
We direct the user to build for infrastructure agnosticism by:
- Creating content that's semantically clear and well-structured so any model architecture can parse, cite, and synthesize it
- Building genuine expertise that transcends platform rather than reverse-engineering what one particular model wants
- Thinking globally about access patterns because AI infrastructure in emerging markets looks fundamentally different
- Embracing citability and transparency because clearly cited, transparently sourced expertise becomes the foundation everything else builds on
Build for Humans, Not Hegemonies
Phil concludes with a warning: the cost of inaction isn't falling behind in a race but becoming irrelevant to the infrastructure layer that shapes global AI adoption. He's right. But I'd frame it differently.
The cost of inaction is becoming irrelevant to the networked, orchestrated, human-driven future that's already emerging. What matters is what users do with the tools they have access to. And what they're doing is brilliant: they're taking the efficiency of Chinese open-weight models, the reasoning capabilities of American frontier models, and orchestrating them together in ways that make questions of national dominance irrelevant.
For content creators thinking about Generative Engine Optimization:
Stop betting on a single platform (build content that remains valuable across the entire ecosystem).
Stop gaming algorithms (become genuinely authoritative enough that every system recognizes your value).
Start building for infrastructure agnosticism (the only content that survives is content that deserves to).
The race isn't to AGI. The race is to build systems, content, and strategies that empower human creativity regardless of geopolitical outcomes. And maybe, in a world where chokepoints are vanishing and users are orchestrating across platforms with incredible ingenuity, that's actually a future worth optimizing for.
Because the future of AI? It's human.
Read Phil Fersht's full analysis: "Why the US is Currently Ceding the AI Arms Race to China"
Learn more about building infrastructure-agnostic content strategies at youcited.com