Peter Thiel's Fund's Single Biggest Reported Position Is Amazon. $10,000 Invested in Amazon 10 Years Ago Is Worth About $66,000 Today.

Amazon generated $717 billion of revenue and $77.7 billion of net income in 2025, versus $136 billion and $2.4 billion, respectively, in 2016; its shares have returned about 555% over the decade. AWS revenue reached $128.7 billion in 2025 with $45.6 billion of operating income and a roughly 35% margin, while Q2 AWS revenue growth accelerated to 36.7% and advertising revenue rose 26% to $19.8 billion. The company is investing heavily in AI infrastructure, contributing to a trailing-12-month free-cash-flow outflow of $7.6 billion, which may pressure margins near term but supports its long-term cloud and AI growth thesis.

Analysis

The relevant investment debate is no longer whether Amazon can grow earnings, but whether incremental AI infrastructure earns above its cost of capital before depreciation and financing burdens reach the income statement. Near-term reported operating income can remain resilient while free-cash-flow conversion deteriorates; this creates a 1-3 quarter risk that the market shifts from rewarding AI capacity buildout to demanding utilization, lease commitments, and depreciation disclosure. The key monitor is not aggregate capex, but AWS growth relative to AWS margin and companywide FCF after equipment purchases.

Amazon has a differentiated internal offset to cloud-capex risk: advertising monetizes high-intent commerce traffic with minimal incremental fulfillment cost, potentially funding part of the infrastructure cycle without requiring retail margin expansion. That makes AMZN structurally less exposed than pure-play cloud vendors to enterprise AI workload delays. Conversely, AWS's custom-silicon push is a second-order risk for NVDA: meaningful adoption of Trainium would pressure the highest-margin portion of accelerator demand, though the effect is likely 6-18 months out rather than an immediate earnings risk.

Consensus is likely too anchored to the prior AWS scaling playbook. Cloud infrastructure was built against a broad migration wave; AI capacity is being installed ahead of less proven application monetization, leaving pricing and utilization more vulnerable if enterprise pilots do not convert. A durable bullish rerating requires evidence that AWS growth remains above 20% while segment margins hold near recent levels and trailing FCF turns positive despite elevated capex; failure on any two of those metrics would challenge the premium multiple.

The disclosed institutional position is not independently actionable: quarterly filings are backward-looking and do not identify hedges, derivatives, or subsequent trading. The actionable catalyst path is Amazon's next two earnings reports, where capex guidance, depreciation trajectory, AWS backlog/conversion commentary, and advertising growth will determine whether investors view spending as productive investment or a margin overhang.

Key Decisions for Investors

• Maintain or initiate a 6-12 month long AMZN position, sized below a normal mega-cap allocation until FCF inflects. Base case is multiple stability with earnings growth; target 15-25% upside if AWS growth sustains above 20% and FCF conversion improves. Reduce if AWS growth falls below 18% for two quarters or segment margin declines more than 300 bps without a clear capacity-driven revenue payoff.
• Use a tactical long AMZN / short NVDA pair over the next 6-18 months only if disclosed Trainium adoption expands materially among external AWS customers. The trade expresses custom-silicon substitution and Amazon's ability to capture AI economics internally; risk is that overall accelerator demand remains supply-constrained enough to overwhelm substitution effects.
• Ahead of the next earnings release, treat a capex increase without explicit utilization, backlog, or revenue-conversion evidence as a risk alert rather than a buy-the-dip signal. A post-results FCF miss combined with higher depreciation guidance could produce a 5-10% multiple reset even if revenue and operating income beat.
• Do not use AAPL or NFLX as read-through trades from this development. Their listed relevance is sentiment-driven rather than tied to Amazon's cloud, advertising, or infrastructure economics.