China, AI Immigration, Rare Earths & Chips, Tariffs, Market Check | BG2 w/ Bill Gurley & Brad Gerstner

Episode Summary

Executive Summary: The discussion compares the internet boom to today’s AI cycle, arguing AI is likely being overestimated short term but underestimated long term. It highlights emerging “data walls,” China’s industrial strategy of seeding massive competition, immigration as a key AI advantage for the U.S., geopolitical risks around rare earths and chips, and market optimism hinging on tariff, tax, and trade deals.

Main Topics: AI vs. the dot-com era (Priority: 5/5): The hosts revisit the late-1990s internet boom and argue AI may follow a similar pattern: slower near-term adoption than expected, but far larger long-term impact than most forecasts assume. Data walls and ecosystem control (Priority: 5/5): They discuss companies like Reddit, Salesforce, and OpenAI as signs that major platforms are beginning to restrict how AI can access and train on proprietary data, signaling a shift toward closed ecosystems. China’s industrial policy and competition model (Priority: 5/5): The conversation unpacks how China seeds industries with hundreds of competitors, then lets market competition and state support determine winners, producing strong outcomes in EVs, robotics, batteries, and AI. Immigration, talent, and U.S. AI leadership (Priority: 5/5): They argue the U.S. should aggressively attract and retain global talent, especially AI researchers and graduates, warning that visa restrictions on Chinese students could damage America’s innovation edge and brand. Rare earths, chips, and geopolitical bargaining (Priority: 5/5): The speakers debate a potential trade: U.S. access to rare earths in exchange for selling China a deprecated NVIDIA chip, arguing it could reduce supply-chain disruption and lower Taiwan escalation risk. Market outlook, tariffs, and fiscal policy (Priority: 4/5): They assess the market rebound and argue future performance depends on tariff outcomes, the reconciliation bill, tax cuts, inflation, and whether the U.S. can avoid a debt-driven slowdown. Delaware and proxy-advisor governance risk (Priority: 4/5): The discussion closes with concern over Delaware incorporation risk and proxy advisory firms ISS and Glass Lewis, which they say may no longer align with shareholder interests and deserve alternatives.

Key Arguments: AI should be viewed like the internet was in 2000: adoption may be slower than expected in the short term, but the long-term economic impact could be vastly larger than current models predict. Companies are increasingly creating 'data walls' to prevent rivals or AI systems from training on their proprietary data, which could fragment the AI ecosystem. China’s success stems less from simple state ownership or IP theft and more from deliberate industrial policy that encourages extreme competition among many entrants before winners emerge. China’s model produces more optionality, stronger supply chains, and cheaper products, which helps explain its strength in EVs, LiDAR, robotics, and open-source AI. The U.S. must preserve open access to talent and data if it wants to win in AI; restrictive visa policies or closed data policies would be strategically damaging. Barring Chinese students and researchers from critical fields risks undermining U.S. national security more than it protects it, because talent is central to AI leadership. A trade of deprecated U.S. chips for Chinese rare earth access could benefit both sides: it would keep China in the NVIDIA ecosystem, generate U.S. revenue, and buy time to build alternative supply chains. Overly aggressive export controls and AI restrictions can backfire by accelerating Chinese self-sufficiency and increasing the risk around Taiwan. The market can still rise if tariff policy settles into a manageable range and fiscal policy extends tax cuts while avoiding an abrupt spending shock. Delaware’s predictability advantage has eroded, so boards should reconsider incorporation location; meanwhile proxy advisors may be too influential and insufficiently aligned with shareholders.

Data Points: Amazon split-adjusted return since 2000 high: ~440x - From the 2000 high ($150/share split-adjusted) to today Amazon split-adjusted return since 2000 low: ~1800x - From the 2000 trough to today Amazon peak share price in 1998: $243/share - Referenced as a dot-com-era high Amazon IPO price: $17/share - It went public and briefly traded below issue NASDAQ return since 2000 peak: 5x - Despite the dot-com bust, the index is still up materially NASDAQ return since 2000 trough: 10x - Long-run gain since the low point in 2000 OpenAI search volume milestone: 400 billion annual searches - Cited as reached eight years faster than Google OpenAI daily searches: Over 1 billion per day - Used to illustrate AI’s rapid adoption AI researchers in the U.S. of Chinese origin: 40% to 50% - Used to argue for welcoming immigration AI patent count in China vs. U.S.: China now larger than the U.S. - Cited as evidence of China’s rise in AI Chinese EV startup count: Over 500 startups - Used to illustrate China’s competitive seeding approach U.S. EV startups considered legitimate: A handful - Contrasted with China’s startup volume Chinese LiDAR car cost: About $130/car - Example of low-cost Chinese solid-state MEMS LiDAR Waymo LiDAR car cost: About $5,000/car - Comparison point for different innovation paths Current U.S. skilled immigration volume: About 200,000 to 250,000/year - Mentioned as too low and stagnant for decades U.S. 10-year Treasury range: 4% to 5% - Described as the recent band, despite alarm over deficits Core PCE direction: Lower than expected - Used to support the case for potential Fed rate cuts Proxy advisors market share: 97% - ISS and Glass Lewis combined influence on corporate voting Proxy advisor ownership: Over 80% owned outside the U.S. - Cited in discussion of their corporate philosophies Delaware fee multipliers: Up to 66x normal hourly rate - From a Fortune-reported study on legal fees Potential leading-edge chip capacity in U.S. by 2030: 15% to 20% - Projected increase in advanced-node domestic production Potential U.S. advanced-node share with UAE joint fab: 40% to 50% in four years - Hypothetical joint-deal scenario discussed

Pivotal Quotes: "If you and I want to win championships and build a championship basketball team, we should not care where in the world the basketball player comes from." — Brad: Argument for prioritizing talent over nationality in AI and technology policy "We ought to take the same approach to AI and technology." — Brad: Extending the sports analogy to immigration and AI hiring "I worry more in general that the China hawk mindset leads you to policy that's really bad." — Brad: Warning against overly hardline China policy that could backfire

Implications: AI leadership will depend on openness: to talent, data, trade, and strategic flexibility. Closed ecosystems, blunt export controls, and anti-immigrant policy may weaken U.S. competitiveness and accelerate China’s rise.