Key Concepts:
- NVIDIA chip export restrictions to China
- U.S.-China trade negotiations and technology competition
- AI development and spending by Chinese tech companies (Tencent, Bytedance)
- Revenue sharing agreements for market access ("pay to play")
- AI infrastructure and GPU demand (training vs. inference)
- Bullish crypto IPO
- SpaceX competition (ULA Vulcan rocket, European launches)
1. NVIDIA Chip Export Restrictions and U.S.-China Relations:
- Treasury Secretary Scott Bessent stated the U.S. is open to discussing the use of NVIDIA chips with China, but will not sell advanced chips (H20s). The goal is to prevent Huawei from dominating technology standards.
- The administration believes lower-performance chips are not a national security concern and wants American tech to maintain a presence in China.
- Export licenses for H20 chips were initially banned, then reversed, becoming part of trade negotiations where the U.S. needed rare earths.
- There is a possibility of discussing "unenhanced Blackwell chips" with China.
- The administration is considering a model where companies share revenue with the U.S. government in exchange for access to the Chinese market, potentially extending beyond NVIDIA.
- The Chinese government may be discouraging domestic firms from using U.S. chips, favoring national champions.
2. AI Development and Spending in China:
- Tencent reported fast growth in gaming and ad businesses, with increased AI spending. They plan to invest "smartly" in AI research and integration across services.
- Tencent's R&D spending was around $19 billion in the quarter, a fraction of U.S. tech giants' spending. They claim to have enough AI chips for now and will focus on recruiting talent.
- Tencent's international games grew by 35%, and ad revenue increased by 20%, benefiting from AI adoption.
- Bytedance is considered a leading AI company.
- Chinese companies are developing their own AI models, with some achieving impressive results at lower costs than U.S. companies.
3. Revenue Sharing Agreements and Market Access:
- The potential for a "pay to play" arrangement where companies share revenue with the U.S. government for market access is being considered.
- President Trump reportedly discussed a 20% revenue share with NVIDIA CEO Jensen Huang, which was negotiated down to 15%.
- This model could potentially be replicated with other chip makers or in other industries.
- The leverage in U.S.-China negotiations will determine the future of these agreements.
4. AI Infrastructure and GPU Demand:
- CoreWeave CEO discussed the massive demand for both AI training and inference.
- Inference demand now accounts for over 50% of their compute usage.
- Newer GPU architectures are used for leading-edge training, while older generations are used for inference.
- There is still strong demand for older GPUs like H100s for inference.
- CoreWeave is diversifying its client base beyond OpenAI and Microsoft, seeing growth in sectors like VFX and life sciences.
- They are open to working with companies like Meta and xAI, believing partnerships are necessary for scaling AI infrastructure.
5. Bullish Crypto IPO:
- Crypto company Bullish raised over $1 billion in an IPO, pricing shares above the market range.
- The IPO was oversubscribed, with strong demand from institutional and retail investors.
- Management had a significant say in share allocation, favoring long-term supporters.
- The IPO valuation is lower than previous SPAC merger attempts.
6. Competition in Space Launch Market:
- ULA launched its Vulcan rocket on its first national security mission, marking a significant step in competing with SpaceX.
- Europe also completed its third mission, aiming to reduce dependency on SpaceX.
- These launches provide more options for satellite providers seeking launch services.
- Both Vulcan and the European rocket are struggling to ramp up their launch cadence.
7. Other Company Earnings and News:
- Shares of 10 Cent are rising after posting quarter revenue, beat missed expectations.
- Cisco is baking in the impact of tariffs into their expectations.
- Amazon will offer same-day delivery in 203 cities this year.
- Soft Phone plans to sale riskier net in an effort to fuel it’s A.I. drive.
- Elon Musk must face claims by OpenAI weaponize social media posts in statements to sabotage OpenAI’s success.
- The billionaire family controlling Porsche and Volkswagen are setting up a new fund on defense investments.
8. Notable Quotes:
- Treasury Secretary Scott Bessent: "We would not sell any of the advanced chips... What we do not want here is for Huawei to have a dej tal belt in the road."
- CoreWeave CEO: "We continue to see massive demand for train. That hasn’t abated in any way... What you are seeing is a continual cranking up of the demand for inference."
- Mitchell Green: "If you could figure out how to read the U.S. and Chinese government relations between what they say and what they do, please keep me enlightened and inform me."
9. Technical Terms:
- H20: A specific type of NVIDIA chip with restricted export to China.
- Blackwell: A newer generation of NVIDIA chips.
- Inference: Using a trained AI model to make predictions or decisions.
- Training: The process of teaching an AI model using large datasets.
- GPU: Graphics Processing Unit, a type of processor well-suited for AI tasks.
- AI: Artificial Intelligence.
- I.P.O.: Initial Public Offering.
10. Synthesis/Conclusion:
The broadcast covers a range of topics related to technology, finance, and geopolitics. Key themes include the complex relationship between the U.S. and China in the technology sector, particularly regarding AI and semiconductor technology. The potential for revenue-sharing agreements as a condition for market access is a significant development. The AI infrastructure boom continues, with strong demand for both training and inference capabilities. Finally, the emergence of competition in the space launch market signals a shift away from SpaceX's dominance.
AI summaries can miss context or contain errors. Check important details against the original video.





