The US is either going to dominate AI or get steamrolled. No middle ground.
That’s the bold claim that hooked me into this whole rabbit hole. 40% of the US stock market sits in seven tech companies, all heavily tied to AI working out. So when something threatens that bet, I pay attention.
I just watched a sharp breakdown from this AI creator, and his core argument is uncomfortable: US open-source AI is almost certainly doomed because there’s no business model that makes it work here. Meanwhile, China is quietly winning the open-source race, and most American businesses are about to bake that into their stack for the next decade.
🔑 The Core Idea
The creator lays it out clean. Open-source means you publish the recipe and the weights, anyone can download it, fine-tune it, run it locally. Llama, Qwen, Gemma, DeepSeek, all open. The problem? In the US, if you spend months and millions baking a model and give it away, your competitors serve it with bigger margins because they skipped the R&D bill. In China, the CCP subsidizes winners and uses free as a strategy to crush leaders. Two very different games.
💡 Three Insights From the Video
- 🔹 The US open-source landscape is collapsing. Meta backed off Llama. OpenAI shipped GPT-OSS as a side quest. Anthropic has zero open-source strategy and is straight-shotting AGI. Google’s Gemma is great but built for local, not frontier. Nvidia might be the only player with the money, talent, and incentive to keep going, because every competitor serving open models is still buying their chips.
- 🔹 Enterprise is choosing Chinese models right now. The author points out that 99% of business use cases are spreadsheets, coding, and scheduling. They don’t need frontier IQ. DeepSeek is nearly as good, a fraction of the cost, runs on your own servers, and is more controllable. Of course companies are picking it.
- 🔹 The geopolitical risk is wild. If US enterprise builds on Chinese open-source, China gets to set AI standards, optimize models for their own chips, and shape cultural defaults baked into the weights. Export controls pushed China to find efficiencies on their own silicon, and that pipeline now runs upstream of everything.
🛠️ Solutions the Creator Floats
- Federal compute quotas or grants for US open-source labs, treating it like public infrastructure
- Tax credits and sovereign procurement in defense, health, finance, energy for US open models
- AMD and Intel copying Nvidia’s playbook: fund open models that run best on your own hardware
- Stop chasing closed frontier models. Build vertical open-source models for legal, biotech, code, defense
- Define open standards early so startups stop reinventing wheels
🤔 The Counterargument
The expert is fair here. He notes Anthropic’s view: only the straight shot to AGI matters. Once one lab hits the recursive self-improving loop, nothing else counts. Maybe true. But nobody knows the timeline, and in the interim China gets to dictate the rails the rest of us build on.
The video closes on the kicker: DeepSeek dropped a new model days before this was filmed and basically proved every point.
👉 Watch the full breakdown for the full landscape map and the policy fixes. It’s one of the clearest takes I’ve seen on why this fight matters.