Prof G Markets — “China’s AI Is 20x Cheaper — And Catching Up”

Date: February 19, 2026
Hosts: Ed Elson (D), with guest Alice Han (F, Host of China Decode), and Robin Brooks (C, Brookings Institution)
Network: Vox Media Podcast Network
Main Theme:
Exploring the rapid rise and strategic shift of Chinese AI models, their global market impact, and implications for U.S. AI companies and enterprise. Includes wider discussion on currency geopolitics and the evolving landscape of AI-government relations in America.

Episode Overview

This episode centers on two main themes:

• The explosive growth and unique market positioning of Chinese AI models, with a focus on affordability, benchmarking, and use cases versus their American rivals.
• Europe’s currency maneuvering in response to shifting global economic power, specifically Sweden’s contemplation of adopting the euro amid global de-dollarization debates.
• Closing commentary on the intensifying Anthropic vs. Pentagon feud, using it as a lens on the future of AI, surveillance, and state power.

1. China’s AI Surge: Cheaper, Accessible, and Competitive

Key Discussion Points & Insights

A. “Lunar New Year: China’s AI Launch Season”

[02:01–03:35]

• Ed Elson outlines the flurry of new Chinese AI releases:
  • Alibaba: Rin Brain (robotics-oriented), Quen 3.5 (coding/agent model, up to 5x faster)
  • ByteDance: Sea Dance 2.0 (video generation), Dubao 2.0 (deep reasoning)
  • Zhipu: GLM5 (agentic intelligence, benchmarking leadership)
• All new Chinese models reportedly matching or surpassing U.S. rivals on key benchmarks.

B. Chinese vs. U.S. AI Models: Use Case & Market Dynamics

[03:39–06:31]

• Alice Han: “Chinese AI has bolted from the stables, to borrow a horse metaphor.”
• Distinct value propositions:
  • Chinese models:
    • Open, highly accessible, 10–20x cheaper than U.S. rivals
    • Favor local downloading, privacy, customization for engineers, startups, and niche projects
    • Designed for experimentation and cost efficiency
  • U.S. models (OpenAI, Anthropic, Claude):
    • Focused on the highest precision, frontier-level reasoning
    • Optimized for enterprise and complex analytics
• Quote:

  "The way that I think about it is that the American models are really focusing on precision, on reasoning at the highest level as a really quest for AGI. The Chinese models are really figuring out how to create a market in which everyday people, engineers, corporates even, are experimenting, fine-tuning, and just capturing the benefits of cheaper, faster models."
  —Alice Han [05:29]

C. Benchmarking: Red Herrings & Market Implications

[06:31–08:30]

• Ed Elson questions the reliability and significance of benchmark leaderboards.
• Alice Han:
  • Benchmarks can be “red herrings”—headline results often don’t reflect deeper capabilities.
  • Real differentiation: cost and ease of adoption; Chinese models are “10, 20 times cheaper” and “inertia will continue to favor the Chinese.”
  • Use cases vary: privacy, language support, and practical deployment trump raw leaderboard stats.

D. Cost Disruption & Market Segmentation

[08:30–10:59]

• Ed wonders when cost will upend Western AI giants; could OpenAI et al. be threatened by models that are not only fast and local, but also a fraction of the cost?
• Alice expects U.S. companies will hold on to large, sticky enterprise clients, but:
  • Chinese models will thrive in non-enterprise markets, new consumer-facing toys, and multilingual domains.
  • Quote:

    “We haven’t even talked about some of the other use cases. There's now Chinese AI chatbot toys... going to be a big market and certainly it marries well with China's manufacturing and hardware capabilities... Another area in which OpenAI and Anthropic have not been as competitive is in multilanguage.”
    —Alice Han [09:58]

E. The State of the Chinese AI Race

[10:59–12:23]

• Deepseek, once seen as a leader, is still formidable but now focused on government partnerships and public governance as AI becomes part of China’s national five-year plan ("AI plus").
• China’s AI sector is not a duopoly; hypercompetition persists among multiple players (Alibaba, ByteDance, Kimmy, Deepseek).
• Quote:

  “In China at least, the AI race is hyper competitive. It’s not just a duopoly... Many different players and I think they can capture different parts of the market.”
  —Alice Han [12:13]

2. The Euro’s Rising Star & The “De-Dollarization” Temptation

Sweden Considers Moving from Krona to Euro

[16:22–26:39]

A. Sweden’s Currency Soul-Search

[17:16–20:09]

• Robin Brooks (Brookings Institution) provides context:

  • Sweden (like UK, Denmark, etc.) opted not to join the euro at its inception; a 2003 referendum rejected it.
  • Now, geopolitics (especially Russia’s regional aggression and U.S. unpredictability) is pushing public opinion toward euro adoption, seeking “safety in numbers.”
  • Brooks dismisses currency bloc membership as a true safety guarantee, calling it “more symbolic than practical.”

• Quote:

  “If that is what’s going on, then I think that’s poor reasoning because the euro is really just a system of currency pegs. It doesn’t confer any kind of safety... do [Estonia, Latvia, Lithuania] feel any safer from Russian invasion because they’re members of the euro? No.”
  —Robin Brooks [18:26]

B. De-Dollarization & Reserve Currency Realities

[20:09–25:13]

• Ed asks if recent euro momentum is part of a broader de-dollarization trend given America’s debt and dollar “hegemony crisis.”
• Robin Brooks:
  • IMF surveys show no substantive shift in global central bank allocations despite recent dollar volatility, gold spikes, and geopolitical drama.
  • Dollar remains dominant; the euro has not absorbed fleeing capital.
  • Sweden’s Krona, counterintuitively, is trading more like a safe haven (akin to the Swiss Franc or gold).
  • “The hurdle for the dollar to lose reserve currency status is pretty high. ...the United States is worrying, but relatively speaking, it doesn’t look that bad.”

C. Final Takeaway

[26:14–26:39]

• Ed: “What’s your alternative? Euro, yuan, bitcoin, gold?”
• Robin: “These are all, all very interesting questions.”
• The episode closes with the acknowledgment that, despite real fiscal issues, U.S. dollar dominance endures for lack of credible alternatives.

3. Anthropic vs. The Pentagon: AI, Surveillance, and State Power

The Feud and Its Implications

[26:44–End]

• Ed Elson recaps the increasingly tense standoff between Anthropic and Secretary of War Pete Hegseth:

  • Pentagon is threatening to blacklist Anthropic over refusal to use its AI for autonomous lethal weapons and domestic mass surveillance.
  • Adding Anthropic to a “supply chain risk list” (typically for foreign adversaries), with secondary boycotts for partner companies.

• Key Issues:

  1. Anthropic refuses to allow AI for weaponized drones or mass surveillance of American citizens.
  2. Pentagon insists on full compliance and is punishing noncompliance.

• Quote (Ed):

  “So there’s a lot that is concerning about this, the fact that we’re now requiring that AI be used for things like drone strikes... also that we’re requiring that AI be used to spy on Americans and surveil them. These facts are scary in and of themselves.” [27:35]
  “What is perhaps more striking is the hypocrisy here... this is exactly the thing that MAGA was supposed to prevent, this collaboration between the government or the deep state and also with technology companies to surveil the American people. And now we find ourselves in this very ironic position where an AI company actually doesn’t want to move in that direction... and for that, they are being punished.” [28:19]

• Advanced Reflection:

  • Most AI companies (OpenAI, XAI, Google, Palantir) are “on board” with weaponization and surveillance.
  • Growing public distrust of AI: less than half of Americans “like” AI, less than a third “trust” it.
  • Potential for public backlash at the voting booth; but unless that boils over, the fusion of AI and government surveillance is inevitable.

4. Notable Quotes & Moments (with Timestamps)

• Alice Han:

  “Chinese AI has bolted from the stables, to borrow a horse metaphor.” [04:11]

• Alice Han:

  “It’s cheaper, often 10 to 20 times cheaper than what’s offered by OpenAI and Claude. So really they’re using this very differently from say potentially you and me, Ed...” [05:07]

• Ed Elson:

  “At a certain point, if you’re presented with an option that is 10 or 20 times cheaper, you go with that option. So I guess what I’m a little confused about is when is that going to happen...?” [08:30]

• Alice Han:

  “We haven’t even talked about... Chinese AI chatbot toys... already living in that diamond age... and then another area in which I would say OpenAI and Anthropic have not been as competitive is multilanguage...” [09:58]

• Robin Brooks:

  “The euro is really just a system of currency pegs. It doesn’t confer any kind of safety.” [18:26] “The hurdle for the dollar to lose reserve currency status is pretty high.” [22:37]

• Ed Elson:

  “The de-dollarization question… has generated a lot of heat… Because it communicates a possibility that this world order… may be coming to an end.” [24:00]

• Ed Elson:

  “We now know where this is all headed. ...AI will be used to track people, target people, and sometimes kill people. Everything that MAGA said that they were supposed to prevent, ultimately that will transpire.” [29:48]

5. Timestamps for Key Segments

| Segment | Timestamp | |-------------------------------------------------|-------------------| | China’s AI launch season and model advances | 02:01 – 06:31 | | Benchmarking and business models | 06:31 – 10:59 | | Deepseek and China’s multifirm AI ecosystem | 10:59 – 12:23 | | Sweden’s euro debate & euro/dollar backdrop | 16:22 – 26:39 | | Anthropic vs. Pentagon: AI ethics/government | 26:44 – End |

Summary & Takeaways

• China’s AI sector is rapidly innovating and claiming benchmarks, with a “faster, cheaper, open” approach, capturing global users and diverse use cases; U.S. models remain strong in enterprise and high-precision analytics.
• Benchmarks can mislead; practical utility, multilingual support, and cost are tilting many real-world use cases toward Chinese platforms.
• Currency strategy highlights growing global uncertainty, but the dollar’s dominance endures—Sweden’s interest in the euro is more symbolic than transformative.
• The Anthropic–Pentagon feud encapsulates the paradox—AI is certain to be bent toward government ends, with privacy and public oversight as the last, uncertain brake.

For more episodes and insights, visit [Prof G Markets].