Thread by @wmhuo168

American AI doesn't stand a chance.

Not because of China’s models, but because our own labs answer to Wall Street instead of Washington.

https://x.com/wmhuo168/status/1938436935694864387?t=LH81u6Qpierx7QJlEoeG9A&s=09

Every American AI lab reports to a stock ticker.

That’s the difference. RAND just showed us what a national AI strategy actually looks like. It’s time to name the problem. Thread 🧵

(1/9)

You want public AI infrastructure? Good luck getting that from Nvidia, Google, or OpenAI.

Their goal isn’t national resilience. It’s quarterly returns. They serve shareholders, not the public. (2/9)

China builds national labs with five-year targets.

We build brand decks for VC pitches. Then get distracted by "safety" panels run by failed philosophers and mid-tier MBAs. (3/9)

They deploy AI in drones, hospitals, power grids, and classrooms.

We build another chatbot and call it "AGI." All while fighting over who gets access to GPUs. (4/9)

RAND makes it clear. China's AI stack is strategic, planned, and treated as national infrastructure.

Our stack is financialized, siloed, and bottlenecked by a handful of rent-seekers in San Jose. (5/9)

We let three public companies monopolize compute. Then act surprised when nobody outside the cartel can train anything.

“Open AI” is open in name only. And it never belonged to the public. (6/9)

It’s not about beating China’s models. It’s about breaking free from a system where AI is whatever makes stock prices go up.

The enemy isn’t Chinese tech. It’s our own captured architecture. (7/9)

This won’t be fixed by more VC. It will only be fixed by public infrastructure, public compute, and a clear mandate to serve the nation, not Nasdaq. (8/9)

RAND gave you the blueprint.

📘

Use it. Before the last sovereign function gets swallowed by API tokens and stock-based comp plans.

(9/9)rand.org/pubs/research_…