Investors should prioritize the hardware layer of the AI boom, specifically targeting undervalued memory stocks like Micron (MU), Samsung, and SK Hynix, which are currently trading at attractive multiples between 5x and 7x earnings. NVIDIA (NVDA) remains a high-conviction play as it powers the "Neo-Cloud" infrastructure led by Elon Musk’s Colossus data center. Monitor Anthropic as the current momentum leader in the AI race, as its massive revenue acceleration and new compute deal with Musk position it to challenge OpenAI for enterprise dominance. The next 500 days represent a critical "net long" window for AI investments before the market demands proof of margin expansion from software applications. To hedge against power bottlenecks, look for companies with secured energy infrastructure and data center capacity, which are becoming the primary gatekeepers of AI scaling.

Elon Musk / SpaceX / xAI (EWS)

The discussion centered on Elon Musk’s rapid build-out of the Colossus data center and the recent deal to lease compute capacity to Anthropic. The hosts dubbed this emerging business "Elon Web Services" (EWS), positioning Musk as a new hyperscaler competing with Amazon (AWS), Microsoft (Azure), and Google (GCP).

• Compute as a Revenue Stream: Musk leased the Colossus 1 data center (containing NVIDIA H100 GPUs) to Anthropic. This provides immediate cash flow to offset the massive CapEx of building these facilities.
• Subsidizing Grok: By leasing out capacity to competitors, Musk can subsidize the training of his own AI, Grok, and the development of more advanced clusters (Macrohard and Macroharder) featuring NVIDIA Blackwell chips.
• SpaceX Valuation: Analysts suggest this "terrestrial" compute business provides a structural core that de-risks the SpaceX valuation as it moves toward an eventual IPO, potentially commanding a premium (40-50x revenue) due to its multi-layered "cake" of launch, connectivity (Starlink), and compute.

Takeaways

• Hyperscale Competition: Musk is no longer just a consumer of chips; he is a provider. This creates a "Neo-Cloud" that competes directly with the MAG7 cloud providers.
• Infrastructure Lead: Musk’s ability to secure power and build data centers faster than traditional tech giants is a significant competitive advantage in a supply-constrained market.

Anthropic

The podcast highlighted Anthropic’s staggering growth and its recent deal with Elon Musk to solve its compute constraints.

• Hyper-Growth: David Sachs noted that Anthropic grew from $10B ARR to $44B ARR in just four months (January to April), a rate of growth rarely seen in Silicon Valley.
• Monopoly Concerns: Sachs argued that if Anthropic maintains its 10x annual growth trajectory, it could reach $100B ARR by the end of the year and potentially $1T ARR by 2027, making it a "monopoly" in the AI space.
• Compute Constraints Solved: The deal for 220,000 NVIDIA GPUs and 300 megawatts of power from Musk has already allowed Anthropic to double rate limits for its Claude models.

Takeaways

• The "AI Race" Leader: While OpenAI was the early favorite, the hosts suggest Anthropic currently has the most momentum in terms of revenue acceleration.
• Enterprise Focus: Much of this growth is driven by "coding tokens"—enterprises using AI to generate and manage software code.

NVIDIA (NVDA) & Memory Stocks

The "Besties" remain highly bullish on the hardware layer of the AI boom, specifically the "memory wall."

• Valuation vs. Growth: Brad Gerstner argued that despite all-time highs, these stocks are not in a bubble. He cited NVIDIA trading at roughly 19x earnings and Meta at 17x, which he considers reasonable given their growth.
• The Memory Play: Gerstner revealed that 25% of his portfolio is in memory stocks, which he believes are undervalued:
  • SK Hynix: Trading at 5x earnings.
  • Samsung: Trading at 6x earnings.
  • Micron (MU): Trading at 7x earnings.

Takeaways

• Hardware is the Foundation: The immediate ROI in AI is currently being captured by the hardware providers (compute and memory) rather than the software application layer.
• Investment Theme: Look beyond the GPU manufacturers to the companies providing the high-bandwidth memory required to run these massive models.

Investment Themes & Sector Insights

The "FDA for AI" and Regulatory Risk

There is significant discussion regarding a potential "vibe shift" in Washington toward regulating AI models.

• Regulatory Capture: The hosts warned that "safety" rhetoric might be used by leading AI firms to create an "FDA for AI," which would act as a barrier to entry for smaller competitors and startups.
• Impact: If a strict approval regime is implemented, it could slow the pace of American innovation and benefit incumbents like OpenAI and Anthropic by "pulling up the ladder" behind them.

Energy and Power Constraints

• The New Gold: Power is the primary bottleneck for AI. Projects are being protested or delayed, making existing "secure power" (like Musk’s data centers) extremely valuable.
• The "Drill, Baby, Drill" for AI: The hosts emphasized that the US needs to unleash energy production (nuclear, gas) to sustain the AI boom and maintain a lead over China.

The "ROI Reckoning"

• 500-Day Window: Chamath Pali-Hapitiya suggested we are in a window (roughly 500 days) where being "net long" AI is the clear play.
• The Fork in the Road: Eventually, the S&P 500 must show measurable margin expansion from AI. If companies spend on "tokens" but don't see increased profits or decreased OpEx, a market correction is likely.

Blue-Collar AI Boom

• Construction & Infrastructure: The AI boom is currently driving a "blue-collar boom" in data center construction, leading to 25-30% wage increases for construction workers in regions like Texas.