The Index
400 dossiers
A research screener for every company we cover. Search a name or ticker, then sort and triage dossiers by coverage freshness, our conviction and trading relevance.
No call7 names· avg TR 2 | ||||||||
| The best business in AI on every metric that isn't valuation — enterprise-led, multi-cloud, ~$30B+ run-rate doubling on weeks — wrapped in a ~$965B mark that already prices a decade of flawless execution and rests on a circular ring of vendor-investors funding ~$180B of compute the company hasn't yet earned the margins to pay for. WATCHING, bull on the business / bear on the entry price. | — | 1 | ||||||
| The shelf RE-GROUNDS and partly CORRECTS the prior web-only read. Three filings-level facts reshape the thesis. (1) Concentration is NOT "OpenAI is the new whale" — as of the reported period the top two customers are MBZUAI (62% of FY25 revenue / 77.9% of AR) and G42 (24%), BOTH UAE-affiliated related parties = 86% of FY25 revenue; OpenAI is a forward 750MW contract (deploys 2026-2028) whose revenue has barely begun. (2) FY25 GAAP "net income $237.8M" is an accounting artifact of +$390.7M of non-operating Other income — dominated by a one-time $363.3M non-cash gain on extinguishment of the Series F forward-contract liability — while operations LOST $145.9M and non-GAAP net loss was $(75.7)M. (3) The margin guide-down is substantially a *reported-revenue* effect (customer-warrant amortization that reduces revenue + datacenter pass-through booked gross + rent-back start-up costs), per management a 10-15pt transitional drag, not purely "inference is structurally low-margin." Net: real wafer-scale architectural edge, genuinely circular AI-infra financing now on the balance sheet (~$1.0B OpenAI working-capital loan at 6% + OpenAI & AWS customer warrants), still UAE-concentrated revenue, ~46x FY26-guide sales vs CoreWeave at ~7x. WATCHING, bear-leaning on valuation; the structural short case is weaker than the prior dossier asserted and the *transitional-margin* read is more defensible. | ||||||||
| The lab that broke the cost-of-intelligence narrative and now anchors China's sovereign AI stack — a frontier-class open-weight model trailing the closed leaders by 3-6 months at ~1/50th the unit cost, but state-captured, un-investable to US capital, and monetizing almost none of the value it creates. | — | 1 | ||||||
G | The crown jewels and the brain already left for Nvidia for ~$17–20B; what remains is a cash-rich but architecturally-cornered inference cloud whose flagship Saudi demand evaporated — a melting-ice-cube neocloud, not the LPU disruptor the headlines sold. | — | 1 | |||||
| Europe's only credible frontier lab — a sovereignty-and-infrastructure bet, not a model-quality bet; the moat is regulatory/political, the risk is that "good enough, in Europe, on-prem" gets out-priced by Chinese open weights before the €20B mark is earned. | — | 1 | ||||||
| The revenue leader that turned into the cash-burn leader — a $1.4T compute book financed against a P&L losing ~$14B a year, where the only question that matters is whether the IPO window opens before the funding treadmill stalls. | — | 1 | ||||||
X | A frontier lab fused into Musk's compute-and-distribution empire — Grok is a fast-following #4 model bankrolled by a $30B/yr capex furnace; the real instrument is now SpaceX equity at a $1.25T mark, not a standalone xAI bet. | — | 1 | |||||
| — |
| 8 |