Tech Brew Ride Home – "China’s Getting Tetchy Again"

Podcast: Tech Brew Ride Home
Host: Brian McCullough (Morning Brew)
Date: October 10, 2025
Episode Theme:
In this episode, Brian McCullough covers rising tensions in the US–China tech trade war, rapid changes in AI development and regulation, the surprising vulnerabilities in AI models, the explosive growth of prediction markets, and EA’s make-or-break moment with Battlefield 6.

Key Topics and Insights

1. China Tightens Tech Trade and Retaliates

• Crackdown on Nvidia Chips ([01:12])
  • China has significantly intensified customs checks on chip imports (especially Nvidia’s H20 and RTX Pro 6000D), aiming to stifle smuggling and enforce US export controls.
  • Chinese regulators are urging firms to avoid Nvidia products.
  • A billion dollars’ worth of Nvidia AI chips were reportedly smuggled in just three months.
  • Quote:

    "Teams of customs officers have been mobilized at major ports across the country in the past few weeks to carry out stringent checks on semiconductor shipments..." – [01:46]

• Ban on Tech Insights ([02:44])
  • China’s Commerce Ministry designated Canadian tech research firm Tech Insights as an “unreliable entity,” banning Chinese organizations from working with them.
  • This move comes after Tech Insights revealed that Huawei’s new AI chips contained components from outside mainland China.
• U.S. Response ([04:21])
  • The U.S. Senate passed a measure requiring Nvidia and AMD to prioritize American customers over Chinese sales of advanced AI chips.

2. OpenAI’s Sora Outpaces ChatGPT’s Growth and Navigates Copyright Backlash

• Rapid Growth and Copyright Concerns ([04:47])
  • Sora reached one million downloads in under five days—faster than ChatGPT did at launch.
  • OpenAI initially put the burden on rights holders to request IP removal but has started limiting character likeness usage after pushback.
  • The Motion Picture Association called on OpenAI to address copyright infringement issues urgently.
    • Quote:

      “Videos that infringe our members, films, shows and characters have proliferated on OpenAI’s service. OpenAI needs to take immediate and decisive action...” — Charles Rifkin, MPA CEO ([05:37])

  • Sam Altman, OpenAI CEO, promised more granular controls for rights holders and asked for patience during rapid changes.
    • Quote:

      “Please give us some grace. The rate of change will be high.” — Sam Altman ([06:25])

3. AI Coding Assistants: OpenAI Catches Up to Anthropic

• Performance Metrics ([06:52])
  • OpenAI’s Codex (GPT-5) now has a 74.3% success rate in code approval, edging past Anthropic’s Claude Code at 73.7%.
  • Adoption and usage still lean toward Anthropic’s offerings (Claude Code usage higher by percentage and by downloads).
  • Largest code acceptance rate (76.8%) is with SourceGraph’s AMP agent, which, despite less hype, is well-funded and competitive.
  • Insight:
    Cost isn’t always the main factor for devs choosing coding assistants; productivity boosts are valued more ([08:56]).
    • Quote:

      “Developers are willing to pay more right now because many of them believe costs will drop over time...” — (Brexton Pham, modu co-founder) ([09:03])

    • For CEOs, “it’s a lot cheaper to pay for coding assistants... than it is to hire more human software engineers.”

4. Meta Pushes for Massive Internal AI Adoption

• Internal Memo from Meta’s VP of Metaverse, Vishal Shah ([10:48])
  • Mandate: Engineers should use AI to work 5x faster, not just 5% faster.
  • Goal: 80% of Metaverse staff using AI in day-to-day work by end of year.
  • Adoption will expand beyond engineers to PMs, designers, and partners for rapid prototyping and innovation.
    • Quote:

      “Make AI a habit, not a novelty. This means prioritizing training and adoption for everyone so that using AI becomes second nature...” — Vishal Shah ([11:26])

    • Future: Feedback loops measured in hours, not weeks.
  • Push includes training and company-wide Day of AI learning events.

5. Surprisingly Easy to Poison an LLM

• Anthropic Study on AI Poisoning ([12:47])
  • As few as 250 malicious documents can produce a backdoor in a large language model, regardless of size.
  • Common assumptions about needing to control a large chunk of training data are false—LLMs are far more vulnerable.
    • Quote:

      “Poisoning AI models might be way easier than previously thought if an Anthropic study is anything to go on.” ([13:10])

  • Recommendations for defense:
    • Continued clean training, enhanced data filtering, and post-training mitigation.
    • The need for scalable defenses, even if the number of poisoned documents is low.
    • Quote:

      “It is important for defenders to not be caught unaware of attacks they thought were impossible…” — Anthropic ([14:08])

6. Prediction Markets Boom: Kalshi Surges to $5 Billion Valuation

• Kalshi’s Explosive Growth ([14:45])
  • Raised $300 million Series D at $5 billion valuation, expanding internationally to 140 countries and $50 billion in projected annualized trading volume (from just $300 million in 2024).
  • Overtook Polymarket, now with over 60% of global activity.
  • Growth fueled by sports contracts, new features (parlays), and integration with platforms like Robinhood and Webull.
  • Faces regulatory scrutiny from several US states over whether financial products tied to sports outcomes are gambling.
  • Quote:

    “Every time there’s a new type of financial innovation, there’s always a series of questions around regulation... If there weren’t questions, what you’re doing is probably not meaningful or innovative enough.” — Tarek Mansour, Kalshi CEO ([16:20])

7. Long Read: The High Stakes of EA’s Battlefield 6 Launch

• Overview from The Verge ([17:01])
  • Battlefield 6’s performance will decide the fate of EA as it prepares to go private in a $55B deal.
  • A flop would force EA to rely more heavily on sports franchises, potentially reshaping the company.
  • After Battlefield 2042’s disastrous launch, Vince Zampella was put in charge to right the franchise.
    • Quote:

      "If Battlefield 6 underperforms, it will call into question the strength of EA’s non-sports portfolio... Battlefield is a bellwether that will tell us if EA can still compete in blockbuster shooters." — Joost van Drunen, NYU Games professor ([17:43])

  • EA’s revenue is already massively skewed toward live services and ongoing monetization.

Notable Quotes & Memorable Moments

• “Teams of customs officers have been mobilized at major ports across the country...to carry out stringent checks on semiconductor shipments.” ([01:46])
• “Videos that infringe our members, films, shows and characters have proliferated on OpenAI’s service. OpenAI needs to take immediate and decisive action...” — Charles Rifkin, MPA CEO ([05:37])
• “Please give us some grace. The rate of change will be high.” — Sam Altman ([06:25])
• “Developers are willing to pay more right now because many of them believe costs will drop over time...” — Brexton Pham, modu ([09:03])
• “Make AI a habit, not a novelty.” — Vishal Shah, Meta ([11:26])
• "Poisoning AI models might be way easier than previously thought..." ([13:10])
• “If there weren’t questions [about regulation], what you’re doing is probably not meaningful or innovative enough.” — Tarek Mansour, Kalshi CEO ([16:20])
• "If Battlefield 6 underperforms, it will call into question the strength of EA’s non-sports portfolio... It makes Battlefield a bellwether..." — Joost van Drunen ([17:43])

Timestamps for Major Segments

• [01:12] China tightens chip import controls, Nvidia crackdown
• [02:44] Ban on Tech Insights, national security concerns
• [04:21] U.S. Senate prioritizes AI chips for domestic use
• [04:47] OpenAI’s Sora rapid growth, copyright pushback
• [06:52] AI coding: OpenAI vs Anthropic performance and adoption
• [10:48] Meta's internal AI adoption mandate
• [12:47] Anthropic: LLMs easily poisoned with minimal malicious data
• [14:45] Kalshi’s funding and prediction markets boom
• [17:01] EA’s future and stakes of Battlefield 6 release

Overall Tone and Language

Brian McCullough keeps the pace snappy, blending clear summaries with direct quotes from primary sources and a touch of cynical humor, especially when noting the industry’s rapid pivots and the high stakes of ongoing tech battles.

For those in tech, this episode delivers a sharp, digestible overview of policy clashes, startup surges, and potentially game-changing vulnerabilities—all with actionable insights and memorable commentary.