fed, carrots, burry, nvidia

Here's a comprehensive summary of the YouTube video transcript:

Key Concepts

• Nvidia's Response to Michael Bur: Nvidia's investor relations team issued a memo addressing allegations made by Michael Bur regarding stock-based compensation, share buybacks, days of sales outstanding, and inventory levels.
• Google's TPU Push: Google is actively pitching its Tensor Processing Units (TPUs) to major companies like Meta and financial institutions as an alternative to Nvidia's GPUs, potentially impacting Nvidia's market share.
• AI Progress and S-Curves: The discussion explores the trajectory of AI development, contrasting a "fairy tale" exponential growth to AGI with a more realistic "S-curve" model where progress eventually plateaus.
• Federal Reserve Policy: An article by Nick T suggests that the Federal Reserve's decision on interest rate cuts will hinge on Chair Powell's assessment of economic crosscurrents, particularly job growth versus inflation.
• ADP Data: The importance of upcoming ADP employment data as a key indicator for the Federal Reserve's rate cut decisions is highlighted.
• Reinvest AI Product: The speaker mentions an upcoming AI product from their company, Reinvest, aiming to accelerate their IPO timeline.

Nvidia's Response to Michael Bur

Nvidia has publicly responded to a seven-page memo from Michael Bur, addressing several of his criticisms.

• Stock-Based Compensation (SBC) and Dilution:

  • Bur's Claim: Nvidia earned $405 billion in net income and $188 billion in free cash flow since 2018. SBC was $20 billion, but with an additional $112 billion in stock buybacks, the "true cost" of SBC dilution was $112 billion, reducing owners' earnings by 50%. This implies 47 million more shares outstanding.
  • Nvidia's Response: Nvidia states they repurchased $91 billion, not $112 billion or $118 billion. They claim Bur incorrectly included RSU taxes and that employee equity grants should not be conflated with the repurchase program. Nvidia argues that employee compensation is not consistent with peers.
  • Analysis: The speaker finds Nvidia's response to be splitting hairs, noting a significant difference in the numbers presented and suggesting that even if taxes are accounted for, the core issue of dilution remains.

• Repurchase Program and Intrinsic Value:

  • Bur's Implication: The argument suggests that by issuing dilutive stock compensation, existing shareholders were deprived of upside as the company evolved.
  • Nvidia's Response: Nvidia claims their repurchases from 2018 were made well below intrinsic value, creating substantial value. They argue that the "intrinsic value" concept is a way of saying the market undervalued them or that business evolved.
  • Analysis: The speaker views this as a petty argument, acknowledging that stock compensation is intended to reward employees.

• Days of Sales Outstanding (DSO):

  • Bur's Claim: Nvidia's DSO increased to 53 days from a historical average of 46 days, suggesting issues collecting revenue.
  • Nvidia's Response: Nvidia states their DSO was 52 days from 2020-2024 fiscal year, not 46 days, and that they are in a similar position. They assert they are not struggling to collect from customers and overdue accounts receivable are negligible.
  • Analysis: Nvidia claims Bur has incorrect numbers and that their situation is stable.

• Inventory Growth:

  • Bur's Claim: Growing inventory in Q3 suggests weak demand, unsold chips, or customers accepting delivery without payment capability, leading to inventory converting to receivables.
  • Nvidia's Response:
    • Growing inventory doesn't necessarily indicate weak demand; it could be due to business growth.
    • Finished goods inventory includes significant raw materials and work in progress.
    • Companies with sophisticated supply chains build inventory in advance of new product launches to avoid stockouts.
    • Nvidia's current supply levels are consistent with historic trends.
    • Nvidia recognizes revenue upon shipping and deeming collectibility probable.
    • Shipments reduce inventory, unrelated to customer payments.
    • Customers undergo strict credit evaluation, and overdue accounts are negligible.
  • Analysis: The speaker criticizes Nvidia's response, noting that it doesn't fully address the significant increase in finished goods inventory (up 109% while raw materials were up 23%). They feel Nvidia punted on key questions and raised more concerns than they answered. The speaker also points out that Nvidia's 10-Q filing showed raw materials up 23%, work-in-progress up 156%, and finished goods up 109% over nine months.

• Comparison to Enron:

  • Nvidia's Statement: Nvidia explicitly states they do not use special purpose entities (SPEs) to hide debt and inflate revenue, unlike Enron.
  • Analysis: The speaker finds this ironic, as Meta is reportedly using SPEs (via Blue Owl) to finance Nvidia data centers, keeping debt off Meta's balance sheet. This is seen as a smart move by Meta but potentially problematic for the industry if Meta pulls back.

• Vendor Financing:

  • Nvidia's Statement: Nvidia states they do not rely on vendor financing arrangements to grow revenue and that customers pay over years with standard payment terms.
  • Analysis: The speaker argues that Nvidia's practice of leasing back or guaranteeing leases on the very products they sell is analogous to vendor financing, as it provides a stable cash flow stream that helps customers secure loans. This is compared to a car dealer artificially inflating a buyer's income to qualify them for a loan.

Google's TPU Push and Market Impact

• Meta's Interest: Meta Platforms is reportedly in talks to spend billions on Google's Tensor Processing Units (TPUs) and may also rent chips from Google's cloud division.
• Google's Strategy: Google is pitching TPUs to customers, including Meta and large financial institutions, as an alternative to Nvidia's GPUs. This strategy aims to boost Google's cloud revenue.
• Potential Market Share: The Information suggests this could potentially capture as much as 10% of Nvidia's annual revenue.
• Financial Impact on Google:
  • Assuming Nvidia's data center sales are around $80 billion annually and Google achieves a 70% gross margin on TPU sales to Meta, this could translate to an additional $5.6 billion in gross profit.
  • With incremental operating expenses assumed to be low (R&D already spent), and accounting for taxes, this could add approximately $4.4 billion in net income.
  • For Google (Alphabet), with roughly 87 million shares outstanding, this could add about $1.1 per quarter to net income, resulting in approximately a 3% increase in EPS.
  • Google's stock saw a significant rise (up 3.44% in after-hours trading) following this news.
• Competition: This move directly challenges Nvidia's dominance in the AI chip market and also impacts AMD, which many hoped would gain traction.
• Google's Other Deals: Google previously secured a deal to supply one million chips to Anthropic, which analysts viewed as a strong validation for TPUs.
• Google's Playbook: Google is also starting to backstop data center leases, similar to Nvidia's strategy, as seen with their guaranteed lease for Fluid Stack's data center.
• Analysis: The speaker views this as a significant development, creating competition and potentially impacting Nvidia's revenue. While the immediate EPS impact on Google is around 3%, the long-term implications and momentum are significant. The speaker questions whether Google is entering the market at an early (A/B) or later (C/D) stage of the AI S-curve.

AI Progress and the S-Curve Theory

• Two Theses:
  1. AI Bull Thesis (Fairy Tale): AI is on an exponential curve leading directly to Artificial General Intelligence (AGI).
  2. Realistic AI Thesis (S-Curve): AI experienced an initial exponential growth phase (step change) but is now transitioning to a logarithmic curve, where progress begins to plateau.
• Implications of the S-Curve: In the later stages of an S-curve, increased investment in compute yields diminishing returns in terms of output or productivity. The speaker likens this to Tesla's Full Self-Driving development, where early progress was rapid, but later improvements are incremental and harder to discern.
• Investment Strategy: The speaker suggests that the best time to invest in AI startups is during the early "A" or "B" phases of the S-curve, when there is significant growth potential. Investing at later stages (C or D) may mean the most significant gains have already occurred.
• Reinvest AI Product Analogy: The speaker uses their upcoming AI product for real estate as an example, aiming to be at an early stage of the S-curve to accelerate their company's IPO.

Federal Reserve Policy and Economic Indicators

• Nick T Article: An article by Nick T in The Wall Street Journal suggests that Federal Reserve Chair Powell's decisions on interest rate cuts will be pivotal due to a divided committee.
• Economic Crosscurrents: The committee is split between stagnating job growth and elevated inflation, creating a risk of stagflation.
• Powell's Dilemma: Powell faces a choice between prioritizing job growth (by cutting rates) or controlling inflation.
• ADP Employment Data: The speaker emphasizes the critical role of upcoming ADP employment data (weekly and monthly) as key indicators that Powell will likely rely on before making rate cut decisions. Three ADP prints are expected before the next FOMC meeting.
• Speaker's Critique: The speaker criticizes Nick T's article for neglecting to mention the significance of the ADP data prints.

Other Noteworthy Points

• Grindr Financing Issue: A proposed $3 billion deal to take Grindr private failed due to uncertainty in financing.
• Quantum Computing: The speaker dismisses "quantum annealing" companies as scams, suggesting that genuine quantum computing investments are with large players like Google or IBM.
• Reinvest AI Product Launch: The speaker is excited about the upcoming launch of their AI product for real estate, aiming for a December release and an IPO between 2027-2030. They are offering lifetime memberships to help accelerate this process.
• Nvidia Stock Performance: Nvidia's stock was down approximately 2% in after-hours trading, while Google was up significantly. The QQQ (Cues) remained relatively flat.
• Gemini's Improvement: The speaker notes that Google's Gemini AI has significantly improved, particularly version 2.5.