The Whole AI Winners Map on One Page
A one-page capstone to the AI market-map series: the companies best positioned to win as models commoditize, usage explodes, and value migrates to infrastructure, distribution, cybersecurity, and scar
How to use this
This is the fourth and final piece in a series that attacked one question from three angles: who actually wins as AI gets commoditized? The three source maps, each in a line:
The Cybersecurity Map (7/17): the sub-sectors of cyber and the best name in each, and why cyber wins on AI diffusion itself: it arms attackers and upgrades defenders whichever model wins.
The Cheap-Model Winners (7/18): who profits when models get cheap and everyone can self-host: the clouds, silicon, data, and app franchises that capture the value as the model layer commoditizes.
When Building Is Easy, Marketing Is the Moat (7/19): as building software goes to near-zero cost, distribution becomes the moat: the marketing and go-to-market names that sell the AI-software flood to paying humans.
Every figure comes straight from those three reports, gathered in one place.
The thread across all three: as intelligence turns into a commodity, the money migrates off the model layer. Down to the infrastructure that meters usage, up to the platforms that distribute it, out to whoever owns scarce human attention. The engine is the Jevons paradox. After the DeepSeek scare, hyperscaler capex didn’t fall; it grew, from ~$410-450B in 2025 toward ~$725B in 2026.
Print this, keep it by the screen, and drill into the source pieces when a name gets interesting.
The master map
Piece key: Cyber = 7/17, Winners = 7/18, Distribution = 7/19. All figures verbatim from the source reports; quarters as labeled there. Each table is ranked strongest-first: the best, most-investible name leads.
Cybersecurity
CRWD, single-agent platform, agentic SOC. +26% (Q1 FY27), accelerating; ARR $5.51B +24%; 34% FCF margin; NRR 115%. Tailwind: Strong. (Cyber)
PANW, the mega-consolidator platform. +31% headline / +14% organic (Q3 FY26); organic NGS ARR +28% to $6.5B; 38.5% TTM adj FCF margin. Tailwind: Strong. (Cyber)
FTNT, the sector’s cash machine. +20% (Q1 26); product revenue +41%; record 36% op margin, $1.01B quarterly FCF. Tailwind: Strong. (Cyber)
ZS, zero-trust perimeter replacement. +25% (Q3 FY26); ARR $3.53B (~21% organic); record 23% op margin; ~6.55x sales. Tailwind: Strong. (Cyber)
RBRK, cyber-resilience via immutable backup. +39% (Q1 FY27), fastest in the map; sub ARR $1.57B +32%; NRR ~120%; FCF $74M. Tailwind: Strong. (Cyber)
OKTA, neutral identity utility. +11% (Q1 FY27); 35% FCF margin; NRR 107%; non-human identities outnumber humans 144:1. Tailwind: Strong. (Cyber)
VRNS, purest DSPM / AI-data governance. +27% (Q1 26); SaaS ARR +69% (+29% ex-conversions); GAAP loss, in transition. Tailwind: Strong. (Cyber)
NTSK, first-tier SSE/SASE, rough finances. +28% (Q1 FY27); ARR $845M +29%; NRR 113%; FCF swung to −$57.2M (−28% margin). Tailwind: Strong. (Cyber)
S, AI-native endpoint challenger. +21% (Q1 FY27); net-new ARR +55%; newly profitable (~10% op guide). Tailwind: Neutral. (Cyber)
TENB, the vulnerability-scanning standard. +9.6% (Q1 26); record uFCF $88.6M (33.8% margin); FY26 guide ~7.4%. Tailwind: Neutral. (Cyber)
QLYS, the capital-efficiency champion. +10% (Q1 26); best-in-class margins; FY26 guide $721-727M. Tailwind: Neutral. (Cyber)
CHKP, legacy firewall, growth laggard. +5% (Q1 26); subscriptions $323M; EPS +13% on high margins. Tailwind: Neutral. (Cyber)
Note: CYBR is delisted; Palo Alto’s acquisition closed February 11, 2026 (final independent ARR $1.44B, +23%). Its machine-identity franchise is now a core Palo Alto pillar: a bigger slice of PANW than a typical bolt-on, but still a fraction of it.
Hyperscale & AI clouds
MSFT, largest enterprise distribution + #2 cloud. +18% (Q3 FY26); FCF $15.8B qtr (−22%, capex-squeezed); 1.6B Windows devices, 400M+ M365 seats. Tailwind: Strong. (Winners)
GOOGL, distribution + its own inference silicon. +22% (Q1 26); FCF $10.1B qtr / $64.4B TTM; ~$460B cloud backlog. Tailwind: Strong. (Winners)
META, distribution + open-weights weapon. +33% (Q1 26); FCF $12.4B qtr; 3.56B daily users, Llama 1B+ downloads. Tailwind: Strong. (Winners)
AMZN, most model-neutral #1 cloud. +17% (Q1 26), AWS +28%; TTM FCF $1.2B (collapsed ~95%); AWS ~29-31% share. Tailwind: Strong. (Winners)
ORCL, wholesale GPU landlord. +21% (Q4 FY26), OCI +93%; FY26 FCF negative $23.7B; RPO $638B (+363%), roughly half OpenAI. Tailwind: Two-sided. (Winners)
CRWV, leveraged GPU landlord. +112% (Q1 26, $2.08B); $740M net loss; $99.4B backlog; Microsoft ~62-71% of revenue. Tailwind: Two-sided. (Winners)
NBIS, leveraged GPU landlord #2. +684% (Q1 26, $399M); adjusted EBITDA positive ($129.5M); 3.5GW of power; $17.4B Microsoft + $27B Meta contracts. Tailwind: Two-sided. (Winners)
Inference silicon / memory / networking
AVGO, custom-ASIC design partner + networking. +48% (Q2 FY26), AI semis +143%; FCF $10.26B qtr (46% margin); ~$73B AI backlog. Tailwind: Strong. (Winners)
NVDA, ~80%-share accelerator leader. +85% (Q1 FY27, $81.6B); FCF $48.6B in a single quarter. Tailwind: Two-sided. (Winners)
MU, HBM memory near-monopoly. Q3 FY26 rev $41.46B (supercycle); record adj FCF $18.3B; HBM sold out for 2026. Tailwind: Strong. (Winners)
ANET, model-indifferent Ethernet toll. +35% (Q1 26); 47.8% non-GAAP op margin; #1 in >10GbE switching. Tailwind: Strong. (Winners)
AMD, credible second-source GPU. +38% (Q1 26); record FCF $2.6B (25% margin); Meta 6GW MI450 deal. Tailwind: Two-sided. (Winners)
MRVL, #2 custom-ASIC house. +28% (Q1 FY27, $2.418B); FY28 revenue target $16.5B. Tailwind: Two-sided. (Winners)
QCOM, on-device silicon without the franchise. −2% (Q2 FY26, $10.6B); auto >$5B run-rate, 1M+ vehicles. Tailwind: Neutral. (Winners)
INTC, the only US company with leading-edge logic fabrication (18A), the domestic AI-silicon sovereignty/onshoring hedge (against TSMC/Taiwan concentration) even if its AI revenue is small today; turnaround. Foundry $5.4B +16% (Q1 26), 18A yields ahead of plan; DCAI +22% to $5.1B; AI rev >$750M and growing; 6th straight beat. Tailwind: Two-sided. (Added post-series)
INTC was added after the original series; its figures come from Intel’s Q1 2026 report (April 24, 2026), not the three source pieces. It earns its spot on the US leading-edge fab (18A) and supply-chain sovereignty, not AI accelerators, where Gaudi lags NVDA/AMD. Ranked last: strategic optionality, not a current winner.
Edge & on-device
AAPL, purest mega-cap edge inference. +17% (Q2 FY26, $111.2B); OCF $28.7B qtr, $62B net cash; 2.5B+ active devices. Tailwind: Strong. (Winners)
NET (also cyber), AI toll road at the edge. +34% (Q1 26, $639.8M), accelerating; FCF $84.1M (13%); NRR 118%; 330+ cities, within 50ms of 95% of humanity. Tailwind: Strong. (Winners + Cyber)
DOCN, self-hosting on-ramp for the long tail. +22% (Q1 26); AI ARR $170M +221%; 650K+ customers; 2027 growth guide raised to 50%+. Tailwind: Strong. (Winners)
FSLY (also cyber), distant-third CDN, real turnaround. +20% (Q1 26, $173M); FCF turned positive ($4.1M); security revenue +32%; ~4.2x forward sales. Tailwind: Neutral. (Cyber)
App-software franchises
INTU, compliance-moat franchise. +10.4% (Q3 FY26); FY26 FCF ~$7.4B; ~100M customers. Tailwind: Strong. (Winners)
NOW, enterprise workflow plumbing. Subscriptions +22% (Q1 26); FCF $1.67B qtr (44% margin); Now Assist target raised to $1.5B. Tailwind: Two-sided. (Winners)
CRM, sales/service SoR; Slack = network moat + agentic surface. +13% (Q1 FY27); FY26 OCF $15B; Agentforce ARR $1.2B +205%. Tailwind: Two-sided. (Winners)
SAP, the ledger agents must write to. Cloud +27% constant-currency (Q1 26); FY26 FCF ~€10B; cloud backlog €21.9B +25%. Tailwind: Two-sided. (Winners)
ADBE, creative-software monopoly at a crossroads. +12.7% (Q2 FY26); TTM FCF ~$10.3B; 32-33M Creative Cloud subs; AI-First ARR passed $500M. Tailwind: Two-sided. (Winners)
WDAY, most seat-exposed franchise. Sub rev guide +14% (FY26, $8.815B); ~$2.65B FY26 FCF; 60%+ of Fortune 500. Tailwind: Two-sided. (Winners)
AI-native / model-agnostic
PLTR, model-swappable ontology layer. +85% (Q1 26, $1.6B); adj FCF $925M (57% margin); Rule of 40 = 145. Tailwind: Strong. (Winners)
APP, AI ad engine printing cash. +59% (Q1 26, $1.84B); FCF $1.29B qtr; 84-85% adjusted EBITDA margin; open SEC investigation. Tailwind: Strong. (Winners)
DUOL, the cheap-AI-to-margin proof case. +27% (Q1 26); FCF $147.8M (50.6% margin); gross margin +190bps explicitly from lower AI costs. Tailwind: Strong. (Winners)
SHOP, agent-commerce merchant rails. +34% (Q1 26); GMV >$100B; FCF $476M (15%); AI-search orders up 13x. Tailwind: Strong. (Winners)
Data / dev / observability
DDOG, cleanest usage toll on AI sprawl. +32% (Q1 26, $1.006B); FCF $289M (29% margin); ARR >$4B. Tailwind: Strong. (Winners)
SNOW, consumption-priced data cloud. +33% (Q1 FY27), re-accelerating; adj FCF $265.5M (19%); NRR 126%; ~50% of customers use AI weekly. Tailwind: Strong. (Winners)
MDB, the AI-app default database. +25% (Q1 FY27); FCF $197.5M qtr; RPO +88%; 19 of the top-20 US banks. Tailwind: Strong. (Winners)
GTLB, AI dev platform, seat-priced. +23% (Q1 FY27); adj FCF $147M (timing-boosted); NRR decelerating 122→117. Tailwind: Two-sided. (Winners)
Distribution / marketing (GTM)
RDDT, double-monetized human data. +69% (Q1 26, $663M); FCF $311M (~47%); 126.8M daily uniques; ~$60M Google + ~$70M OpenAI data deals. Tailwind: Strong. (Distribution)
HUBS, the SMB platformization winner (one-roof GTM). +23% (Q1 26, $881M); FCF $154M (17%); 299,458 customers; outcome-based agent pricing. Tailwind: Strong. (Distribution)
KVYO, Shopify-native e-commerce marketing. +28% (Q1 26, $358M); FCF $212M TTM; 196,000+ customers; first GAAP profit. Tailwind: Strong. (Distribution)
TWLO, the last-mile pipe to humans. +20% reported / +16% organic (Q1 26, $1.407B), fastest organic since 2022; FCF $132M (~9%); 10M+ developers. Tailwind: Strong. (Distribution)
ZETA, AI agents that run campaigns. +50% (Q1 26, $396.3M), accelerating; 19th straight beat-and-raise; FCF $41.7M (10.5%); GAAP-unprofitable. Tailwind: Strong. (Distribution)
BRZE, enterprise engagement, agentic push. +30% (Q1 FY27), fourth straight accelerating quarter; record FCF $27M (~13%); GAAP-unprofitable. Tailwind: Strong. (Distribution)
TTD, independent ad-buying platform. +12% (Q1 26, $689M), decelerating from +26% FY24; FCF $276M (~40%); 95% client retention. Tailwind: Two-sided. (Distribution)
AMPL, product analytics where AI is a cost center. +17% (Q1 26, $93.5M); FCF negative $13.2M and worsening on AI-inference costs. Tailwind: Two-sided. (Distribution)
SEMR, invented AEO (AI-answer visibility). +18% (FY25, $443.6M); AI ARR $4M→$38M in 2025. Acquired by Adobe. (Distribution)
RAMP, identity / data clean rooms. +9% (FY26, $813M); 20% TTM FCF margin. Being acquired by Publicis. (Distribution)
The distilled verdict
Own under almost any outcome (thesis-robust: they win whichever model, lab, or nation prevails):
Consumption tolls: DDOG, SNOW, ANET, MU
Edge / on-device: AAPL
Model-agnostic: PLTR
Cybersecurity, the cleanest non-binary trade: CRWD, PANW, FTNT, OKTA (wins on AI diffusion whichever model wins)
Distribution scarcity: RDDT
SMB platformization: HUBS (the one-roof consolidation winner)
Why they win: none need to pick the AI winner. They get paid on usage, diffusion, or attention regardless of who does.
Read the label precisely: “almost any outcome” means robust to the model race, not to everything. It is a claim about business durability, not entry price. PLTR sits here and in “thesis pre-paid” below, a franchise can be right on the thesis and still punish you on valuation.
Most innovative:
PLTR, the ontology: the enterprise as a model-swappable API. Nobody else productized it.
GOOGL, TPU/Ironwood: the only hyperscaler whose inference costs fall on its own silicon curve.
NET, AI Gateway + Monetization Gateway: building the toll booths and payment rails of the agent economy before it fully exists.
META, open-weights-as-weapon: giving away a billion-download model family to burn the rivals’ business model.
RDDT, monetizing scarce authentic human signal twice: ads plus AI data licensing, while AI answers cite it rather than replace it.
OKTA, extending identity to the machine: governing the non-human identities (AI agents, service accounts, API keys) that now outnumber humans 144:1, the fastest-growing gap in security.
Two-sided / debated:
Seat-priced software (CRM, NOW, WDAY, GTLB): the precision matters. The risk is per-tool seats and the per-seat pricing model, not total headcount (automation grows total employment). The tell: vendors are abandoning per-seat for consumption pricing. NOW is best-positioned (tenant-level consumption pools); WDAY weakest (its bear case rests on total headcount falling).
Thesis pre-paid: PLTR at 75-90x forward earnings (~450% premium), NET at ~28-35x sales, NVDA with 80%+ embedded growth. Right on fundamentals and painful at the entry price are compatible outcomes.
The model labs themselves: the commoditized middle. Chinese models undercut Western API pricing by up to 9x; the blunt phrase is “margins disappear.” Not directly tradeable, but the economics leak into everyone contracted to them (see ORCL’s OpenAI-half backlog and its BBB− downgrade).
The three through-lines
Jevons wins: cheaper AI expands usage faster than it cuts prices. Capex grew after DeepSeek. Own the meters, not the commodity.
Cyber doesn’t need a winner: AI diffusion arms attackers and funds defenders at once. It’s the one AI trade agnostic to the outcome.
When everyone can build, nobody can be found: distribution and attention are the moats software can no longer be.
A discipline note on the semis correction (SOX −20%, in an active drawdown): Paul Tudor Jones kept three words above his desk: “Losers average losers.” A sector in an active downtrend isn’t a bargain until the selling exhausts and it bases. The fundamentals above pick the names; the price action picks the moment. We don’t handicap which comes first.
Bottom line
The whole series compresses to one idea: as intelligence becomes a commodity ingredient, own the tolls (the meters that bill usage: chips, memory, networking, data, monitoring), the storefronts (the distribution franchises that serve AI to installed bases), and the scarce human attention (the machinery of being found). Not the commodity in the middle, where margins go to disappear.
📊 Read this on SignalDeck.live.
Not financial advice. For informational and educational purposes only. Do your own research.