NVIDIA generated $216 billion in revenue with 75% gross margins — the most dominant operating performance in tech history. But the FY2026 10-K reveals $165 billion in total obligations growing 2-5x faster than revenue, a $17 billion Groq license that created $14.4 billion in goodwill on zero revenue, and a $40 billion equity portfolio entangling NVIDIA with its own customer ecosystem. These aren't reasons to sell — they're the specific conditions investors should monitor at 25x forward earnings.
Palantir paid $23 million in taxes on $1.66 billion in pretax income — a 1.37% effective rate. The FY2025 10-K reveals exactly why: a single line item in the tax footnote worth $720 million that rises and falls with the stock price. That mechanism boosted reported net income by 19.4%, putting the real P/E closer to 247x than the reported 199x. The filing also shows that only 37% of the $11.2 billion remaining deal value is contractually binding, and that international commercial revenue grew just 2.4% in a year the company grew 56%. The business transformation is real — but so is the amplification.
Palantir (14.6% SBC) and AppLovin (2.4% SBC) both grew revenue ~65% in Q3 2025. So why does one require 6x more equity compensation? Our SEC filing analysis reveals diverging trajectories that matter more than static spreads: APP's SBC is declining 38% YoY while PLTR's is accelerating 42% YoY. The difference: human-intensive vs AI-engine scaling economics.
Two enterprise AI giants, two very different earnings quality profiles. Palantir (6/10) wins on cash conversion, accounting practices, and litigation risk. Snowflake (4/10) struggles with negative cash conversion (-0.47x) and aggressive software capitalization ($228M). Our 5-pass SEC filing analysis reveals what standard screeners miss.
AVGO's top 5 customers account for 40% of revenue. NVDA's largest customer is 22%. ANET depends on Microsoft (20%) and Meta (15%). Standard AI stock screens show none of this. Here's our 2-signal framework for stress-testing AI beneficiaries.