“...pleasure and pain, they come and go and don't last... bear them patiently”—Gita
Each name links to a full research note built from SEC filings and fund documents, with every figure sourced and every valuation assumption stated. Listed by evidence strength: strongest-supported notes first.
Search, YouTube and Google Cloud: Q2 2026 revenue up 24% to $119.8 billion, Cloud up 82%, but 2026 capex of $195–205 billion has turned quarterly free cash flow negative. On our best-estimate assumptions (20–25× exit P/E), the base case implies about 12% a year (8% to 16%).
Custom AI chips plus VMware software: Q3 FY2026 revenue up 86% to $29.6 billion, with AI chip revenue up 221%. On our best-estimate assumptions (18–24× exit P/E), the base case implies about 15% a year (8% to 22%).
The leading-edge foundry: Q2 2026 revenue up 34% to US$40.2 billion at a 67.7% gross margin. On our best-estimate assumptions (15–20× exit P/E), the base case implies about 9% a year (4% to 14%).
AI accelerator leader: fiscal 2026 revenue up 65% to $215.9 billion at a 71% gross margin. On our best-estimate assumptions (18–25× exit P/E), the base case implies about 14% a year (7% to 21%).
Database company turned AI-infrastructure builder: $664 billion of contracted backlog but −$28.7 billion of trailing free cash flow. On our best-estimate assumptions (14–20× exit P/E), the base case implies about 19% a year (12% to 25%).
AI-driven mobile advertising platform: Q2 2026 revenue up 53% at a 78% operating margin, yet the shares are down 69% from their high on fears that growth has peaked. On our best-estimate assumptions (16–22× exit P/E), the base case implies about 20% a year (14% to 27%).
Memory maker at the peak of an AI/HBM upcycle: fiscal 2026 revenue of $133.2 billion at a 74.6% operating margin (unaudited). On our best-estimate assumptions (mid-cycle EPS of 40–60% of the FY2028 consensus, 10–14× exit P/E), the base case implies about 7% a year (−1% to 15%).
AI server builder whose revenue grew 78% to $39.1 billion in fiscal 2026 while free cash flow was −$7.0 billion. On our best-estimate assumptions (10–14× exit P/E), the base case implies about 21% a year (15% to 28%).
NAND flash pure play spun off from Western Digital in February 2025: fiscal 2026 revenue up 175% to $20.2 billion after three loss-making years. On our best-estimate assumptions (mid-cycle EPS of 30–50% of the FY2028 consensus, 9–13× exit P/E), the base case implies about 1% a year (−8% to 9%).
GPU cloud for AI: $104 billion of backlog from Microsoft, OpenAI, Meta and others, financed with $35.6 billion of debt; the shares fell by half in a year as bond yields rose to about 15%. With losses expected through 2027 we use a revenue multiple (3.0–4.5× 2031 sales, less projected net debt); the base case implies about 15% a year (−13% to 43%).
Listed in June 2026 at $135 (the ticker once belonged to a SPAC ETF). Starlink, launch and AI at a $1.83 trillion market value. With EPS near breakeven we use a revenue multiple (8–12× 2031 sales); the base case implies about −7% a year (−18% to 3%).
Concentrated, leveraged growth fund with a 1.91% expense ratio: SpaceX is 35.6% and Tesla 14.8% of net assets. Returned 25.2% a year over ten years. Our look-through projection (SpaceX on our revenue-multiple range, Tesla on a stated return range, the rest on earnings) implies about 2% a year in the base case (−4% to 8%).