All Research
Page 9 of 10 · 219 reports
AI in cars has crossed from demo-grade into revenue-grade, but the revenue structure is sharply tiered. The first to monetize are L2/L2+ ADAS and the "pick-and-shovel" suppliers—GM Super Cruise (FY26 guidance near $400 million, roughly 70% gross margin), Qualcomm automotive (Q2 $1.3 billion, +38%), Mobileye (2025 $1.894 billion, +15%); Robotaxi is genuinely paid but still in the early-scaling phase (Waymo / Apollo Go / Pony.ai / WeRide). Rating Watch: over the short-to-medium term, profit is more likely to flow to ADAS fitment, in-vehicle AI chips/domain controllers, and the few fleet platforms that have already closed the loop.
The enterprise Agent moat does not sit in the model; it sits in "permissions-aware data access + business-system entry points + approval/audit/observability." The financial signals already on the ground: Microsoft's AI annualized revenue tops 37 billion, commercial cRPO reaches 627 billion; ServiceNow's 2026 AI contract ACV could break 1.5 billion, with Now Assist large customers up 130% year over year; Palantir Q1 2026 total revenue +85%, U.S. commercial +133%; Salesforce Agentforce has commoditized its pricing across $2 per conversation, Flex Credits, and $5 per seat. Gartner expects 40% of Agent projects to be cancelled over unclear ROI, with "Agent washing" clearly visible. Key names to track: Microsoft / Salesforce / ServiceNow / Palantir / Atlassian / HubSpot / Workday / Google Cloud. Rating Watch: revenue proof is concentrating in a few platform-layer leaders, while adoption pace and ROI clarity—not model capability—remain the swing factors for the rest.
AI advertising and marketing automation has moved from "showcasing model capabilities" to a "revenue-execution layer." The first profit pool still belongs to traffic-gateway platforms: Google AI Max lifts conversions 14%, Meta's AI ad infrastructure runs at an annualized rate above $60 billion, Amazon Ads totaled roughly $68.6 billion for full-year 2025, plus AppLovin Axon and The Trade Desk Kokai at 5x ROAS. The second profit pool belongs to software platforms that own first-party data, the content supply chain, and workflows: Salesforce Agentforce ARR of $800 million (Agentforce + Data 360 above $2.9 billion), HubSpot Q1 2026 revenue of $881 million, Klaviyo NRR of 110%, Adobe FY26 Q1 AI-first ARR up more than 3x year over year. On the China side, Baidu's AI-native marketing reached ¥2.8 billion in Q3 2025, up 262%, and ¥2.7 billion in Q4, up 110%; Alibaba's "Quanzhantui" penetration is lifting customer-management revenue. On the agency and data side, Publicis acquired Lotame, covering 1.6 billion IDs across 100+ data sources. Standalone, single-point AI creative, AI SDR, and pure generation tools detached from data and channels are the most exposed to being made free by platforms or eroded by price wars. Rating Overweight: the durable profit pools sit with platforms and data-rich software, not with point tools.
An evolution from a small-rocket maker into a Launch + Spacecraft + Critical Components + Mission Solutions platform. 2025 revenue was 602 million, with Space Systems at 67%; Q1 2026 revenue reached 200 million at a 38.2% GAAP gross margin, and backlog hit 2.2 billion (+20%). Yet 2025 operating cash flow was -166 million, CapEx 156 million, and free cash flow deeply negative. At $124.77, the diluted market cap of roughly $78 billion sits far above even the optimistic range, leaving no margin of safety. Rating Watch: a genuinely capable space platform priced for a far-off, very large success, with Neutron's mass production and unit economics the decisive variable.
Palantir has shifted from "high growth but chronic losses" to "high growth plus strong cash flow": FY2025 revenue of 4.475 billion (+56%), GAAP net income to shareholders of 1.625 billion, and free cash flow of 2.101 billion; Q1 2026 revenue grew 85%, with U.S. commercial up 133%. Yet at the current 133.99, a 344.5 billion market cap implies a trailing PE of roughly 150.6x, P/FCF of roughly 128x, and SBC-adjusted P/OE of roughly 176x. This is an excellent company priced for near-perfection, worth owning for the long run but not worth acquiring at today's price. The three DCF scenarios land at 18-30 / 35-60 / 75-110, leaving the current price well above even the optimistic range. Rating Watch: a high-quality compounder whose price already prices in a decade of extraordinary success.
A two-sided Payments + Marketplace + Fintech super-app in Kazakhstan, with 2025 net income attributable to shareholders of KZT 1,073.18 billion, a Tier 1 capital ratio of 19.6%, and a Fitch BBB- rating. At the current $87.78 and a market cap of roughly $16.66 billion, the static P/E is about 7.8x and P/Owner Earnings about 10x, a clear discount to Nu/MELI/Block; that discount reflects Kazakh/Turkish sovereign and regulatory exposure, Hepsiburada integration execution, and governance risk. Rating Watch: a good business at a price that is not cheap.
A basket of U.S. innovation leaders (roughly 3,351 constituents; ONEQ samples 1,026 of them); highly concentrated (top ten ~57.81%, IT + Communications + Consumer Discretionary ~77.56%) with a low cash-distribution rate (30-day SEC yield 0.41%). IXIC currently sits at 26,307 / ONEQ at $103.23, TTM P/E 28.93x, with an apparent earnings yield of 3.46% below the 10-year Treasury's 4.59%. A great asset, but not a cheap price, and the margin of safety is faint. Rating Watch: better suited to long-term growth and index investors than to deep-value buyers.
Ethereum carries roughly half of all stablecoins, 44.1 billion in DeFi TVL, and 33.7 billion in rollup-secured assets. Yet supply has still grown a net 1.177 million ETH since the Merge, an annualized +0.266%, and value capture is weak. At a current price of $2,184, ETH already sits inside the optimistic band and needs a strong narrative to deliver. Rating Watch: an important crypto-finance backbone trading at a fair-to-rich price, not a value opportunity with a margin of safety.
A high-engagement learning platform with an asset-light model, strong cash flow, and a net-cash balance sheet (cash and investments of $1.253 billion in 2026 Q1). But low switching costs, free substitutes, and AI reshaping the boundaries of learning keep the moat only moderate. At $112.06, the headline PE is distorted by a one-time $256.7 million tax benefit booked in 2025. Ideal buy at $85-100, acceptable to hold at $100-125, clearly expensive above $165. Rating Watch: a high-quality business whose moat is not yet wide enough to buy aggressively without a clear margin of safety.
A powerful scarce asset, but not a strong cash-flow asset: protocol-native owner earnings are roughly zero, and today's $77,993 price needs an ever-rising monetary premium to hold up, leaving no margin of safety. It resembles a high-volatility, no-yield bet on scarcity more than a traditional value investment. Research rating Watch: understandable but not cheap, better observed than bought on classic value logic.
BNB is not Binance equity—no dividends, no buybacks, no liquidation rights; value capture works indirectly through "supply-shrinking burns + ecosystem demand." At roughly $645 and a market cap near $87 billion, BSC's P/F of about 778x sits well above Solana (276x) and Tron (91x); the burn-implied annualized "buyback-like yield" of about 4.4% does not clearly beat the 10Y Treasury at 4.63%. Rating Watch: a quality ecosystem asset whose current price offers no meaningful margin of safety—ideal entry $300-450, acceptable $450-650, overvalued above $800.
The real bottleneck in AI chip supply is the compound scarcity of high-end HBM, CoWoS / large-format advanced packaging, N3 / HBM base die, and test/probe capacity. The most certain revenue and margin upside sits with HBM makers, TSMC's 3DFabric, and test and hybrid-bonding equipment leaders. Rating Watch: track TSMC, SK hynix, Micron, ASE, Advantest, BESI, Samsung, Broadcom, FormFactor, and Unimicron, while treating ABF / glass substrate / broad AI PCB names as sympathy plays that still need orders and disclosed breakouts.
Within AI optical communications, the themes that have already converted into orders and revenue rank, in order, as 800G pluggables, early-stage 1.6T ramp, AI switches and switch chips, EML/CW lasers, and key optical components and high-end optical manufacturing; CPO/NPO/Optical I/O are mid-term architectural directions whose contribution to financials over the next twelve months remains weaker than 800G/1.6T pluggables. Priority names to track are Arista, Cisco, Broadcom, Marvell, Coherent, Lumentum, Fabrinet, Innolight, Eoptolink, T&S Communications, and Accelink. Rating Watch: the demand is real, but the order in which profit is distributed across the chain will keep shifting, so the highest-conviction value sits with switch chips, key lasers, and the system software layer rather than with commoditized module assembly.
The second landing point of AI CapEx is shifting from a chip-centric model toward a rack-and-facility-centric one: GPUs and HBM carry the largest value but their profit leverage is already priced in, while order leverage and the profit pool are moving toward rack-scale integration, liquid-cooling core components (CDU, cold plate, QD), power distribution (transformers, PDU, UPS), high-power connectors, and short-reach copper cables. Priority names to track are Vertiv, Eaton, Schneider, Amphenol, HPE, Supermicro, Wiwynn, Quanta, Hon Hai, and Delta. Rating Watch: durable order and profit leverage is migrating from the chip layer to the boundary between the system layer and the facility layer.
The AI bottleneck is shifting from "compute" to "data supply, data movement, and data governance." The segments with the greatest direct revenue sensitivity: enterprise SSD / high-capacity HDD, AI storage systems, consumption-based lakehouse revenue, and search / retrieval / data streaming / governance subscriptions. Priority names to track: Microsoft / Oracle / Micron / Dell / MongoDB / Elastic / WD / Seagate / Palantir / Montage Technology; among private companies, watch Databricks / VAST Data / WEKA / MinIO / Qdrant. Rating Watch: a real, monetizable demand shift whose winners are concentrated in storage hardware under tight supply and platform software with sticky governance moats.
The AI application layer is moving from "feature enhancement" toward "standalone revenue." The tracks that have already proven a revenue model: AI coding (GitHub Copilot/Cursor), AI customer service (NICE/Five9/Intercom Fin), legal and tax verticals (Thomson Reuters/RELX/Wolters Kluwer), and vertical healthcare plus developer tools. Business models are bifurcating across three billing layers: seat + usage + outcome. Rating Watch: prioritize researching Microsoft / ServiceNow / Salesforce / Intuit / Oracle / Thomson Reuters / Wolters Kluwer / NICE / GitLab / Tempus.
Agents are moving from "answering" to "acting"—becoming the enterprise system-of-action execution layer. The clearest revenue evidence on the ground is concentrated in Salesforce Agentforce (ARR 800 million, +169%), Microsoft 365 Copilot (20 million paid seats), NICE AI ARR +66%, HubSpot outcome-based pricing, Palantir Q1 +85%, and UiPath ARR 1.853 billion. The hardest layer to displace is not the model itself but permissions, the workflow entry point, connectors, governance, and distribution. Key names to track: Microsoft / Salesforce / ServiceNow / NICE / HubSpot / Palantir / UiPath / Pegasystems / Appian / SAP. Rating Watch: revenue proof is concentrating in a handful of platform and application-layer leaders, while adoption pace—not model capability—remains the swing factor for the rest.
A great asset at a bad price. At roughly $422 today, the stock has already prepaid far too much for large-scale success in Robotaxi, FSD, and Optimus, leaving no margin of safety. Rating Avoid: a wonderful company, but the price has run far ahead of any cash it can plausibly return; fair buy range is about $70–140 per share.
Great asset, bad price. At roughly $422 today, the stock has already prepaid far too much for large-scale success in Robotaxi, FSD, and Optimus, and carries no margin of safety; a reasonable buy range is about $70-140 per share. Rating Avoid: a strong business whose price demands outcomes too extreme to underwrite as value investing.
A high-quality platform business spanning the WeChat ecosystem, gaming, fintech, cloud, and an investment portfolio, with real cash flow, a robust balance sheet, and a deep moat; but the business is not simple, and at the current price the margin of safety is not obvious. Research rating Hold: a top-tier compounder trading from fair to slightly attractive, not a clearly undervalued bargain.
A superb business at a price that is far from cheap: a $1.25-1.75 trillion valuation already prices in Starship's long-term economics and industrial dominance. The merger with xAI raises the capital-spending burden and governance risk, and a dual-class structure leaves minority shareholders with little say over capital allocation. Rating Watch: a great company at a price that demands a flawless future.
A basket of large U.S. public companies (503 constituents, ~80% of investable market cap) that returned 1.685 trillion dollars in total shareholder payouts over the trailing twelve months. At ~29x Owner Earnings it is not cheap, and future returns hinge on whether the underlying companies can sustain mid-to-high single-digit nominal growth while the valuation anchor holds. Rating Hold: a high-quality core asset at a fair-to-rich price, not a deeply discounted bargain.
A high-quality pharma business: ROIC sits at a lofty 39-88%, with 55% share in once-weekly injectable GLP-1. But facing Lilly's oral GLP-1 launch, U.S. pricing policy, and semaglutide losing exclusivity in some international markets, the stock trades at 20-23x on an Owner Earnings basis. Rating Watch: the cheapness of a static 12.5x PE is an illusion, and today's price offers no clear margin of safety.
Paying a high price for an exceptional business. NVIDIA is now a full-stack platform for AI infrastructure, and its multi-layered moat (the CUDA ecosystem, NVLink system integration, and developer mindshare) is still widening; but at the current ~$225 it trades at 46x PE with a 1.8% FCF yield, the fair buy range is $120 to $160, and the price already sits near the upper edge of the optimistic scenario. Rating Watch: an outstanding franchise priced for sustained perfection rather than a bargain with a thick margin of safety.