Nvidia (NVDA) Investment Plan

Here is the synthesis of the debate regarding NVIDIA (NVDA), viewed through the lens of the Graham-and-Doddsville philosophy.

Debate Summary#

The Bull presents NVIDIA as the "Standard Oil of the digital age," arguing that its CUDA software ecosystem creates an insurmountable moat that locks in 5 million developers. They highlight the company's staggering efficiency—112% ROE and 55% Net Margins—and view the valuation as reasonable because earnings growth is outpacing the stock price. They believe "Sovereign AI" (nations building their own clouds) will extend the growth runway beyond just Big Tech.

The Bear counters that at a $4.5 Trillion market cap, the mathematics of generating a return become prohibitive (the "Law of Large Numbers"). They argue NVIDIA is selling cyclical capital equipment, not recurring consumer goods, and faces a "CapEx digestion" phase similar to Cisco in 2000. Crucially, they point out that NVIDIA’s biggest customers (Microsoft, Google, Amazon) are actively building competing chips to bypass NVIDIA’s pricing, and highlight the geopolitical risk of relying entirely on TSMC—a risk that previously caused Buffett to exit TSMC.

Balanced Takeaway#

NVIDIA is undeniably a "wonderful business" with a fortress balance sheet and a visionary owner-operator, satisfying our quality criteria. However, the price implies a permanent suspension of economic gravity. The Bull’s reliance on "Sovereign AI" and perpetual margin expansion ignores the historical cyclicality of the semiconductor industry. While the moat is wide today, the Bear’s point about "customer-competitors" suggests the moat may narrow just as the valuation demands it to widen. We cannot pay a premium for perfection in a cyclical industry.

Final Stance: WATCH

Investor Commentary#

Core Takeaway NVIDIA represents the classic dilemma of a "wonderful company at a dangerous price." While the business quality is elite, the current market capitalization leaves absolutely no margin of safety. We will admire the fortress from a distance but keep our powder dry for a cyclical downturn or a market dislocation that aligns price with intrinsic value.

Rationale Applying our framework, I cannot justify an allocation at these levels for three specific reasons:

1. The "Price vs. Value" Disconnect: As the Bear noted, a $4.5 trillion valuation requires the company to add hundreds of billions in value annually just to generate a modest return. My previous mistake was failing to analyze fundamental data in the context of market conditions—specifically, assuming that current growth rates can be extrapolated indefinitely. In reality, trees do not grow to the sky. Paying 46x earnings for a hardware company at the peak of a cycle is a recipe for permanent capital impairment.
2. The Customer-Competitor Dynamic: A core tenet of our philosophy is the durability of the "moat." When your largest customers (the hyperscalers) are incentivized by your 76% gross margins to build their own alternatives, your pricing power is on a ticking clock. We prefer businesses like Coca-Cola or Apple, where the customer has no desire or ability to substitute the product.
3. Geopolitical Fragility: We prioritize certainty. The fact that NVIDIA’s entire supply chain relies on a single island (Taiwan)—a risk that forced us to sell TSMC previously—makes this "too hard" of a pile to place a massive bet on at 38x book value.

Monitoring Plan We will add NVDA to our "Wish List" and monitor the following over the next 1-4 quarters: * Gross Margin Compression: Watch for any tick down from the ~75% level. This is the canary in the coal mine indicating that hyperscalers are successfully pushing back on pricing or switching to internal chips (Maia, Trainium). * Hyperscaler CapEx Guidance: Analyze the fundamental data from Amazon, Microsoft, and Meta. If their CapEx growth flattens or declines (digestion phase), NVDA’s earnings will re-rate violently. * The "Rubin" Launch: Watch for execution slips. With a 1-year product cycle, any delay in the next architecture will prick the perfection bubble.

We are disciplined buyers. We will wait for the "AI Winter" narrative to drive the price down to a level where we are paid to take the risk, rather than paying to participate in the euphoria.