Nov 1, 2026 · 60 min
Open models sharpen America’s AI race with China
The American Startup Taking On China’s AI Models
The episode examines whether accessible AI can strengthen American competition or undermine the business models of leading closed-system companies.
- 1Open-weight models offer customization and transparency that closed systems cannot provide through rental-style access.
- 2DeepSeek has intensified debate over AI economics, Chinese technological influence, and the strategic value of open models.
- 3Reflection is targeting enterprise and public-sector customers with American open models designed to address trust concerns.
Don't miss
Laskin argues that banning Chinese open models could hurt American innovation by forcing startups and researchers to depend on incumbent AI labs.
The brief
Ed Elson speaks with Misha Laskin, co-founder of Reflection AI, about the US-China AI race and why cheaper open-weight models are changing the competitive landscape.
Laskin separates open source, open models, and open weights, showing how licensing and transparency determine whether users can inspect, customize, or build on AI systems.
DeepSeek becomes the episode’s central test: Chinese open models may pressure US companies economically, but restricting them could also weaken American startups and researchers.
The conversation challenges the idea that AI will have one winning business model, arguing that open builders, distributors, applications, and closed providers can coexist.
Laskin outlines Reflection’s strategy for enterprise, public-sector, and sovereign customers, then argues that entrepreneurs should pursue problems they understand deeply rather than chase status.
What was said on this episode
21 statements · 11 positive · 5 negative · 1 mixed · 4 neutral
Western open models significantly lagged behind Chinese open models.
“the Western open models started significantly lagging behind China”
Listen at 8:49
America should have the best open-source AI software and models.
“it's extremely important for America as a country to be competitive and not just competitive, but have the best open source software and models when it comes to AI”
Listen at 10:31
Closed AI systems are rented, while open systems can be owned entirely.
“Closed, you cannot own. You can only rent. Open, you can own entirely.”
Listen at 11:04
Open AI models are not necessarily free.
“it doesn't necessarily mean that open means free”
Listen at 13:50
Open models allow architecture inspection, fine-tuning, and customization.
“With an open model, you can. You can see the architecture. You actually can plug it into either your favorite. open source fine tuning framework, load it up on your GPUs and actually customize it for your needs.”
Listen at 22:15
Open-source software tends to be more hardened and robust than closed software.
“open source software tends to be much more hardened and robust than closed software”
Listen at 24:18
More than 90% of the world's servers run open-source operating systems, mostly Linux.
“over 90% of the world's servers, I think the number is even higher, over 95%, run on, you know, some variant of... an open source operating system, most often Linux”
Listen at 26:24
AI models should be treated as commodities like electricity or water.
“the models are a commodity like electricity or water”
Listen at 27:15
AI intelligence per watt will continue increasing.
“the amount of intelligence you'll get per watt is going to keep increasing”
Listen at 29:24
AI's desirable long-term end state is utility-like commodity service.
“the desirable end state for AI is that it's a utility”
Listen at 31:25
Chinese government support likely underpins the Chinese open-model ecosystem.
“it's very hard to explain the Chinese open model ecosystem without... deep support from the government in some way”
Listen at 33:46
Chinese model companies may operate closed APIs domestically while distributing models freely abroad.
“I suspect that, you know, the way in which like their businesses operate within China is, you know, not too dissimilar from a closed API effectively. And then you kind of give away the model, you know, for free abroad.”
Listen at 35:41
Building a model also creates a state-of-the-art serving, agent, and fine-tuning stack.
“by building the model, you've also built Frontier state-of-the-art kind of stack for serving the model, for building agents around it, fine-tuning, and so forth”
Listen at 36:48
Great open models benefit everyone regardless of origin.
“The fact that there are great open models out there. period, doesn't matter where they come from, is a gift to us all”
Listen at 42:10
Lower AI costs have so far increased demand for AI systems.
“when you take AI and you so far, every time you've kind of driven down the costs, the demand has gone up for them”
Listen at 43:58
Chinese open models rarely enter production at Fortune 500 companies.
“It's very rare to see Chinese open models go into production in Fortune 500.”
Listen at 45:21
Allied nations will become dependent on Chinese AI infrastructure.
“other governments, allied nations, will start getting locked into Chinese infrastructure”
Listen at 47:22
Eliminating open source would seriously damage the American economy.
“the lack of open source period. wherever it comes from, will actually be more devastating for like that will be the thing that's devastating for our economy”
Listen at 49:26
Without open source, American startups could not build intelligence or compete with incumbents.
“our startups can't build. their own intelligence and can't compete right with uh you know with large incumbents”
Listen at 49:36
Good open models increase and diversify demand for compute.
“by nature of having a good open model, there's just going to be more demand for, you know, diverse demand for compute”
Listen at 50:44
Open-model and closed-model companies can both succeed.
“there will be open model companies and kind of builders on top of open models and distributors that are successful. Right. And there will be closed model companies that are successful”
Listen at 53:11
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.