← All Reports

The Quiet Alpha Map

What 80 leading venture firms are betting on for the AI era — and where the smart money quietly disagrees.

📅 June 5, 2026 🔭 Galileo Research

Executive Summary

We read 906 forward-looking market claims published over the last twelve months by 80 of the most credible venture firms — from Sequoia, a16z, Founders Fund and Bessemer to specialist voices like Paradigm, NfX, Emergence and Pebblebed — and mapped where they converge and where they break apart.

The method is deliberately simple: a claim's signal is inversely proportional to how many firms already hold it. We sort every thesis by distinct-firm convergence into five bands — from Consensus (priced-in orthodoxy) down to Lone Signals (a single firm willing to say it out loud). The crowded top tells you what's already in the water. The sparse bottom is where edge, if it exists, is hiding.

Three things stand out. First, the consensus has hardened around atoms: robotics, defense, healthcare, reindustrialization — the field has collectively decided the next decade is physical. Second, a quiet but rapidly-forming bet that the SaaS business model itself is breaking — outcome pricing, services-heavy delivery, and "labor-budget" economics replacing the seat-based playbook. Third, the most interesting disagreements are no longer about whether AI matters but about where the value settles: apps vs. infrastructure, open vs. closed, capability vs. trust — and whether the whole thing is mistimed.

The five-second version:

“Maps to” sector key — pills are grouped into colour families:
Digital / AIAgentsAI appsAI labsSoftwarePhysical / atomsRoboticsEnergyDefenseNatural ResourcesHealthHealthcareMoney / market / metaFintechConsumerServicesMacroVC modelMoatsCryptoCryptoUnmatchedOther
Auto-derived from each claim by keyword — a glance aid for cross-referencing sectors, not a ranking.
① Consensus Forming — the priced-in orthodoxy (26+ firms)

What the field has collectively decided. These are real and large — but if you're hearing it from 30 firms, it's in the price. Useful as the baseline to measure everything else against.

ThemeFirmsThe claim, distilled — and why it mattersMaps toVoiced by
Embodied AI is the next platform 37 The single most crowded bet in venture. Robotics, humanoids, and "physical AI" are treated as the inevitable second act once language is solved. The framing has shifted from if to which form factor — and the capital is already moving. Robotics a16z1, Khosla2, Eniac3, 8VC4, Conviction5, DCVC6, Riot7, Fusion Fund8, General Catalyst9, Bessemer10, 809011
Healthcare is AI's first trillion-dollar vertical 34 Near-universal conviction that healthcare's labor budget — not its software budget — is the prize. The disagreement is only on the wedge: clinical documentation, diagnostics, or owning the delivery entity outright. SoftwareHealthcare a16z12, General Catalyst13, Bessemer14, Thrive15, Conviction5, DCVC16, Initialized17
Incumbents win / the labs move up-stack 34 A sober counter to startup romanticism: distribution, data, and capital advantage the incumbents, while frontier labs march up the stack into applications. The consensus is that the model layer is not where most startups should plant a flag. AI labsMoats Sequoia18, Greylock19, SPC20, Spark21, Soma22, Thrive23, Index24
Software is commoditizing — value moves up-stack 32 If a competent model can generate the interface on demand, the UI moat erodes. The crowded conclusion: defensibility migrates to workflow ownership, proprietary context, and the business model — not the software itself. SoftwareAI appsAI labsMoats Slow25, SPC26, Floodgate27, FirstRound28, CRV29, NfX30, Headline31
Defense and dual-use go mainstream 30 American Dynamism graduated from thesis to consensus. Counter-UAS, autonomy, and a software-defined industrial base are now table-stakes venture categories rather than contrarian bets. SoftwareDefense Founders Fund32, 8VC33, a16z1, Riot7, DCVC16, General Catalyst34, 809011
Consumer AI: proactive, voice-first, context-aware 30 The "death of the prompt box": consumer AI that observes, anticipates, and acts rather than waiting to be asked. Passive context capture is the agreed-upon unlock; the open question is who owns the context. Consumer a16z1, Forerunner35, Maveron36, USV37, Khosla38, SPC39, Greylock40
Coding is the killer app 29 The one application everyone agrees has crossed the threshold. The frontier has moved from autocomplete to autonomous background agents — and coding is the proving ground for everything agentic that follows. AgentsAI apps Redpoint41, Sequoia42, a16z1, SPC39, Greylock19, Conviction5, Index43
Agent infrastructure is the new middleware 28 Picks-and-shovels for agents: orchestration, eval, memory, auth, observability. The bet is that today's agents are fragile, and the durable opportunity is the unglamorous tooling that makes them reliable in production. SoftwareAgents Redpoint44, WorkBench45, Bessemer46, NfX47, Greylock48, SPC26, Paradigm49
Reindustrialization and the industrial base 26 Onshoring, supply-chain sovereignty, and the "machine that makes the machine." A macro-political tailwind that has become a default lens for hard-tech investing across the field. DefenseNatural ResourcesMacro Founders Fund32, 8VC4, 809011, DCVC16, General Catalyst34, Eniac50
Energy is the binding constraint on AI 26 Compute is gated by power. The consensus has fully absorbed that data-center demand collides with grid reality, making generation, transmission, and efficiency first-order AI investments rather than adjacent ones. Energy DCVC16, Khosla38, 8VC4, Fusion Fund8, 809051, General Catalyst52
② Interesting — conviction forming, not yet crowded (16–25 firms)

Enough independent firms to be more than noise, few enough that the trade isn't done. This is where a thesis is becoming actionable before it becomes obvious.

ThemeFirmsThe claim, distilled — and why it mattersMaps toVoiced by
Outcome pricing is killing the seat-based model 12 A fast-forming and falsifiable bet: AI monetization shifts from per-seat to output- and outcome-based pricing, dragging gross margins from 80-90% toward 50-60%. The contrarian wrinkle (WorkBench, via Ev Randle): stop valuing these companies on margin and value them on absolute gross-profit dollars per customer — a labor-budget TAM that can be 4-5x a SaaS seat. "AI is worse than SaaS" may be measuring the wrong thing. Software Bessemer53, FounderCollective54, WorkBench55, Emergence56, General Catalyst57, Contrary58, CRV59, Notable60, Felicis61, a16z12, Maveron62, Index63
Proprietary data and feedback loops are the real moat 21 As models commoditize, the defensible asset is data nobody else has. Greylock sharpens it: not input data (commoditizing) but human-in-the-loop feedback — which outputs users accept, reject, or override — producing a hard-to-copy "reasoning graph" of how an organization thinks. Moats Greylock19, a16z64, Emergence65, BasisSet66, General Catalyst52, Initialized67, Index68
The realness premium: AI floods content, authenticity gets scarce 8 If generation cost goes to zero, the scarce good is what can't be generated: live sports, real community, verified-human creativity. A consumer-side scarcity thesis that inverts the abundance narrative — and quietly predicts a backlash market. Consumer Lightspeed69, USV70, IAVentures71, Maveron62, Haystack72, Slow25, Forerunner35, DayOne73
Talent and judgment become the scarce inputs 19 As AI commoditizes execution, the bottleneck moves to taste, judgment, and the rare people who have them. Floodgate's framing: AI "empowers order-makers and threatens order-takers." Felicis extends it to investing itself — when the herd outsources thinking to LLMs, contrarian human judgment appreciates. ConsumerMoats CRV74, Haystack72, Floodgate75, Felicis76, Spark77, SVAngel78, NfX79
This is a bubble — and it will reprice 19 A surprisingly broad skeptic caucus. Redpoint's analogy is the sharpest: AI adoption is moving an order of magnitude faster than cloud, which makes it look more like the internet era — complete with Cisco-style "air pockets" where eyeballs don't convert to revenue, and where today's leaders may not be the eventual giants. Macro Thiel Capital80, Contrary81, Anthology82, HustleFund83, BasisSet66, Slow25, Redpoint44
Capital is commoditized; judgment and trust are the VC moat 10 The field turning the lens on itself. Haystack states it flatly: capital is no longer a differentiator; trust, listening, and selection are. A reflexive admission that the mega-fund "capital-as-a-weapon" playbook may be self-defeating in a world of abundant money. VC modelMoats CRV84, Haystack85, Abstract86, BoxGroup87, IAVentures88, Caffeinated89, SVAngel78, Spark77, 500Global90, Browder91
Agentic commerce and machine-speed payments 9 A standards war nobody's pricing: UCP vs. ACP vs. x402, with TenOneTen betting the real volume is unglamorous B2B back-office execution (logistics, procurement, construction) — not consumer checkout — and that the "protocol-agnostic Switzerland" of agent payments captures value precisely because the standards stay fragmented. CryptoFintechConsumerAgents TenOneTen92, USV93, Paradigm94, Pebblebed95, Redpoint96, General Catalyst52, Interplay97, Browder98, Sequoia99
Non-AI fundamentals still win deals 16 A quiet contrarian steadiness: distribution, retention, unit economics and survival still decide outcomes. Haystack's two-rubric market — "native AI" companies judged on speed vs. everyone else needing an explicit reason to exist — is the sharpest articulation. Moats Haystack72, FounderCollective100, Eniac3, BoxGroup87, First Round101, Cowboy102, Susa103
Inference is the runtime; value migrates to deployment 15 The bet that the action moves from training to serving: edge deployment, KV-cache economics, throughput-first clouds, and small specialized models that out-economize frontier models on unit cost. 8090's version: value accrues to whoever deploys, not whoever builds. AI appsAI labs 809051, Fusion Fund8, Emergence104, Initialized105, BasisSet106, Redpoint44, BOND107
Agents as doers, not copilots 15 The copilot-to-autopilot migration as the real value frontier. General Catalyst's framing: outcome-delivery builds a learning-by-doing flywheel that point-solutions can't match — defensibility lives in accumulated execution data, not the model. AgentsMoats General Catalyst13, Emergence65, SPC20, NfX30, USV93, Sequoia42, WorkBench45
③ Novel — early clusters worth tracking (10–15 firms)

Emerging structure. A handful of credible firms converging on something the broader field hasn't named yet.

ThemeFirmsThe claim, distilled — and why it mattersMaps toVoiced by
The GTM is going services-heavy (FDE / outcome delivery) 8 Forward-deployed engineers, owned service entities, AI-native BPOs. FirstRound's data: FDE job listings up 800% in nine months because capability is no longer the constraint — steering and last-mile integration are. The next enterprise winners may look services-heavy and "un-SaaS-like." SoftwareServices FirstRound28, Index63, Emergence108, General Catalyst13, Interplay97, Headline109, Contrary58, 8090110
Legacy experience is now a liability 6 Sharper than "young founders are cool": several firms argue domain experience is actively disadvantageous in a regime-change market, because incumbic intuition mis-predicts a world that no longer follows the old rules. AStar is moving allocation toward founders in their teens. VC model AStar111, 1517112, Browder91, Slow113, Neo114, BasisSet66
The self-improving, agent-native firm 7 Org design as the product. Sequoia's "$0-to-$1B-revenue club" of companies running agents internally for legal, recruiting and sales; Pebblebed running its own fund as a tech company. The bet: headcount decouples from revenue, and the firm itself becomes software. SoftwareAgentsVC model Pebblebed95, BasisSet115, NfX116, Sequoia18, USV37, FounderCollective117, Greylock48
Inference chips and compute move to the edge 7 On-device, low-power, close-to-the-data. A bet against the cloud-hyperscaler-only narrative, favoring embedded inference and efficient silicon — with YC taking the extreme tail (chips in space). Other Fusion Fund8, 809011, Initialized118, BasisSet119, Redpoint120, Emergence104, YC121
The open web is dying → a closed, licensed, agent-readable internet 6 Flybridge's leading indicator: scrape-to-referral ratios of 2,000:1 (OpenAI) and 71,000:1 (Anthropic) are collapsing the ad/SEO-funded link economy. The next internet is monetized through data licensing, navigated by agents, and optimized for "machine legibility" rather than human eyes. Agents Flybridge122, Lightspeed123, a16z12, USV124, Khosla125, Maveron126
Brand, taste and trust as the durable moat 19 As capability converges across models, the differentiator becomes intangible: brand, taste, emotional attachment. Spark's "muse vs. faster-horse" thesis — products that raise the ceiling earn loyalty; commodity efficiency wrappers don't — is the cleanest statement of where consumer defensibility goes. ConsumerMoats Lightspeed127, Spark128, Maveron129, Forerunner130, Index131, CRV132, Felicis61
Biology is the next AI substrate 20 AI-x-bio as a generational platform: protein design, AI-run wet labs, and an "AlphaFold moment" for engineering simulation (Conviction). Crowded enough to be real, novel enough that the winning form factor is still contested. Healthcare Conviction5, DCVC6, Khosla38, Pebblebed133, a16z12, Initialized134
Seed and stage compression / the barbell 13 Series A bifurcating into $50M "native-AI" rounds vs. $10M everything-else (Haystack). The middle hollows out; capital barbells toward category-defining bets and lean survivors, with the old "raise lean and predict the next round" framework breaking. MacroVC model Haystack135, Pear136, Initialized134, NfX137, BoxGroup87, Conviction138, Abstract86
Permanent capital, rollups and AI-enabled buyouts 13 Buy the boring business, install AI, expand the margin. The PE-venture convergence: own-and-operate service businesses and roll up fragmented verticals where AI is the operational alpha rather than a product feature. Services General Catalyst57, Thrive139, 8VC33, Khosla125, DCVC16, Conviction5
Emerging markets and international as the under-priced lane 10 While US capital crowds into horizontal AI, several firms point at lower-competition international and vertical markets. HustleFund's steer: durable returns come from verticals with workflow moats and geographies with far less competition. SoftwareAI appsMacroMoats HustleFund83, Endeavor140, 500Global90, Tuesday141, Goodwater142, Pantera143
④ Sharp — the almost-nobody bets (2–9 firms)

Two or three serious firms looking at something everyone else is ignoring. Highest risk, highest potential edge — and where the most falsifiable, specific claims live.

ThemeFirmsThe claim, distilled — and why it mattersMaps toVoiced by
Adoption is plateauing — the hype is mistimed 6 The bear spine. Contrary cites usage slipping 46%→37% and 42% of enterprise AI initiatives killed; Anthology measures real productivity gains at 0.6-1.2pp; HustleFund puts a 1-2 year clock on the apps gold rush before a sudden white-collar shock. The dissent isn't "AI is fake" — it's "the timing and magnitude are wrong." AI appsMacro Contrary81, Anthology144, HustleFund83, Redpoint44, BasisSet66, Thiel Capital80
78% of AI failures are invisible and model-immune 2 Susa's striking claim: most production AI failures generate no complaint, no rating, no escalation — and ~94% would persist even with a much better model, because they're driven by interaction dynamics, not capability. If true, "scale fixes everything" is wrong, and an entire AI-supervision/observability category is needed now. Software Susa145, WorkBench146
Real-time compliance becomes its own software category 3 Embedding enforcement into the workflow — validate-as-you-act, auto-log, escalate only edge cases — rather than auditing after the fact. A specific, less-crowded bet tied directly to governing autonomous agent actions. SoftwareAgentsAI apps WorkBench147, Susa148, TenOneTen92
Prediction markets / radical financialization as civic infrastructure 4 Paradigm's political-index work and Slow's "fork in the road" both bet that prediction markets either get regulated into the ground or explode into core infrastructure for expressing views. A genuinely two-sided, high-variance macro bet. FintechMacro Paradigm49, Slow25, Pantera149, DayOne150
Agent identity, KYA, and the security of autonomous action 6 As agents transact, "know-your-agent" and spend controls (session caps, transaction limits, human escalation) become the differentiator. TenOneTen predicts an embarrassing agent-spend blow-up sorts the market — safety rails as moat, not feature. AgentsMoats TenOneTen92, WorkBench147, Pebblebed151, Susa152, a16z153, Browder154
Skilled trades and the physical-labor premium 6 As knowledge work commoditizes, the un-automatable physical trades gain economic and cultural value. An under-explored corollary of the embodiment thesis — the humans who work with atoms get more valuable, not less. Robotics a16z64, 8VC4, DayOne155, Eniac3, Khosla156, DCVC16
Science institutions are broken — fund the heterodox 9 A conviction that academia and grant science are structurally failing, opening room for privately-funded, heterodox research (focused research orgs, independent labs). 1517 takes it furthest — risk culture itself, not capital, is the binding constraint on breakthroughs. Other 1517157, DayOne158, Conviction5, Initialized134, Khosla38, DCVC16
AI-vs-AI commerce: machines selling to machines 9 When the buyer is an agent, the entire funnel — marketing, SEO, pricing, negotiation — re-architects for a non-human customer. Maveron: "ask-and-decide" agents kill the consumer funnel. A structural reframe most commerce companies haven't priced. ConsumerAgents TenOneTen92, USV93, Paradigm94, Maveron126, Redpoint96, Browder98, Sequoia42
Permitting and interconnection as the real energy bottleneck 4 Not generation — connection. The unsexy regulatory and grid-interconnection queue is the actual constraint on powering AI, making the investable opportunity political and infrastructural rather than purely technological. Energy DCVC16, 8VC33, Khosla38, General Catalyst52
Domestic resource extraction and material sovereignty 3 USV's domestic-brine lithium (extract without new mining), plus a broader bet on securing battery-material and critical-mineral supply chains onshore. A specific, physical, supply-chain wager riding the reindustrialization wave. DefenseNatural Resources USV159, 809011, DCVC16
⑤ Lone Signals — a single firm willing to say it (1 firm)

One firm, on the record, against the grain. Most will be wrong. The ones that aren't are where outsized returns come from.

ThemeFirmsThe claim, distilled — and why it mattersMaps toVoiced by
AI breaks GDP and dissolves mineral scarcity Khosla Khosla's maximalist tail: AI-driven productivity makes conventional GDP measurement meaningless and pushes toward $1M-per-capita output, while advanced extraction dissolves the notion of mineral scarcity entirely. Civilizational abundance as an investable base case. Natural ResourcesMacro Khosla38
"Stagnation still holds — AI alone won't end it" Thiel Capital Against the entire productivity-supercycle narrative: outside the world of bits, progress remains stuck. AI is one fast vertical in an otherwise stagnant landscape — and the revealed-preference proof is the firm's reported full Nvidia exit into mega-cap diversification. Macro Thiel Capital80
Venture capital for people, not companies Slow Slow's signature heterodoxy: finance individuals directly — a percentage of a person's 30-year future earnings or a holding company over their output — rather than single companies. Legally novel, productized into a dedicated Creator Fund. VC model Slow160
Restaurants are mis-priced entertainment-fintech assets USV A genuinely strange reframe: restaurants as premium entertainment venues that will support native loyalty and payments layers — paired with USV's decade-old bet that credit-card interchange economics finally die. FintechConsumer USV161
Engineering simulation's "AlphaFold moment" Conviction A single-firm conviction that physical-world simulation (fluids, structures, materials) is about to undergo the same step-change AlphaFold gave biology — collapsing R&D cycles across every atoms-based industry. Healthcare Conviction5
AI inside legacy physical infrastructure — for safety, not speed Initialized Embedding AI into airspace, the grid, and critical systems where the value is reliability and safety, not throughput — an inversion of the move-fast framing in the highest-stakes physical domains. Energy Initialized118
Gen Alpha rewrites consumer: "Sephora Kids" and shitposting-as-marketing Maveron Maveron's bet that Gen Alpha's behavior — early brand fluency, performance-native culture, shitposting as legitimate performance marketing — reshapes consumer company-building well before the cohort has income. ConsumerMoats Maveron62
The deflationary wealth-transfer — short friction and junk fees Browder Browder frames AI as a deflationary force that strips margin from friction-based and junk-fee businesses, transferring wealth to consumers — an explicitly short thesis on the businesses that monetize hassle. FintechConsumerMacro Browder162
A "company brain" skills file as the missing automation layer YC YC's primitive: every company needs a living, executable skills file that extracts scattered domain knowledge — distinct from enterprise search or RAG — as the precondition for reliable business automation. AI apps YC121
Cryonics, longevity and culture as adoption infrastructure DayOne DayOne's tail bets: cryonics as a fundable category, and art/culture as the real infrastructure of technology adoption — the soft layer that determines whether hard tech gets used at all. HealthcareMacro DayOne150
⚔️ Sharp Contradictions — where serious firms directly disagree

The highest-value reading in the report. When credible firms reach opposite conclusions from the same evidence, the disagreement itself is the signal — it marks the questions still genuinely open.

  1. Is the forward-deployed-engineer model the future or a trap? FirstRound calls FDE the emerging default GTM (listings +800%). Flybridge calls it an expensive mistake for 95%+ of startups that destroys unit economics. Both are looking at the same motion and reaching opposite conclusions.
  2. Does value accrue to apps or to infrastructure? Floodgate and Fusion Fund bet value migrates to the application/intent layer (the "Netscapes and Nokias" of infra lose). Spark bets the opposite — concentrate into the deepest infra layers (models, datacenters, chips, lithography), where a few platforms absorb most of the value.
  3. Is capital a weapon or a commodity? Haystack sees $50M native-AI rounds as mega-funds using capital as an offensive weapon. Also Haystack (and CRV) argue capital is now a non-differentiator and judgment is the scarce edge. The same firms hold both — a real tension about whether money still buys advantage.
  4. Open or closed? USV and a16z (crypto) bet permissionless, composable protocols beat closed frontier labs over time. The Incumbents-win consensus bets the opposite. This is the defining structural disagreement of the cycle.
  5. Is adoption inexorable or plateauing? The Consensus bands assume relentless adoption. Contrary, Anthology, and HustleFund present data that usage is slipping, productivity gains are small, and a timing air-pocket is near. Both can't be right about 2026-27.
🛸 Off the Deep End — bets from the far tail

Pulled out for their own spotlight: the strangest single-firm theses in the corpus. We make no claim they're right. We note only that genuinely non-consensus returns, by definition, start out sounding like these.

The betFirmWhy it earns a spotlightMaps to
Inference chips deployed in spaceYC121Cheap reusable launch (SpaceX, Stoke) creates a market for inference compute in orbit — optimized for mass, thermal dissipation, and radiation tolerance. Solar power is free and cooling is the void. The most literally far-out bet in the corpus, and a pure strangeness signal.Other
"Yes, we are at risk of apocalypse — and that is good"Founders Fund163Founders Fund's accelerationist tail: catastrophic-risk technologies are inseparable from their inverse upside — climate doom implies weather control, pandemic risk implies nanobot immunity, AGI risk implies abundance. The rational posture is to accelerate into apocalyptic-grade technology. About as far from mainstream risk-management as a thesis gets.AI labs
P≠NP imposes a permanent ceiling on AIPebblebed164A structural anti-doom argument from computational complexity: if P≠NP, there are hard limits no amount of scaling transcends, bounding what AI can ever do. A rare thesis that uses theoretical computer science to argue the superintelligence runaway is mathematically constrained.AI labs
A live pegasus and de-extinction as commercial categoriesBoostVC165BoostVC's synthetic-biology tail: biology used to make materials that never existed, and genetically-enhanced animal species — explicitly naming a live pegasus and de-extinction — as commercializable venture categories. The literal end of the spectrum where engineered organisms become products, not metaphors.Healthcare
In-space manufacturing and asteroid-belt energy are seed-stage todayCaffeinated89Not a 2050 bet — Caffeinated argues microgravity manufacturing (Varda) and energy ‘abundant from Earth to the asteroid belt’ (Antares) are category-defining seed/Series A markets to lead now. The strangeness isn’t the destination, it’s the claimed timing: orbital industry as a current venture stage, not a moonshot.EnergyMacroVC model
‘The next 100 years will run on nuclear’BoostVC166A century-scale energy call with a near-term trigger: BoostVC bets the first new US reactor design in 50+ years (Radiant’s portable micro-reactor) goes live in 2026, and that portable fission — not solar, not fusion — is the default power source of the next century. Unusual for the explicit 100-year framing and the single-company leading indicator.Energy
Progress is bottlenecked by the rate of ‘better explanations’Floodgate167A Deutsch/Popperian epistemology applied to venture: civilizational and economic progress is gated not by capital or compute but by knowledge-creation — the rate at which humans generate better explanations. The investable corollary: value accrues to whoever can generate and act on new explanations fastest. Philosophy-of-science as an investment thesis.Other
‘Opportunity markets’ — private prediction markets for scoutsParadigm168Paradigm’s new financial primitive: private prediction markets where scouts who spot unsigned bands, breakthrough research, or athletic talent get paid by institutions that can act — with prices kept secret during an ‘opportunity window.’ A genuinely novel market structure that doesn’t map onto any existing asset class.Fintech
Methodology

We harvested 906 forward-looking market claims from 80 venture firms — their essays, theses, annual reports, partner Substacks, and podcast transcripts — published primarily between June 2025 and June 2026. A small number of standing signature theses outside that window are retained and noted where a firm still actively acts on them.

Themes were derived bottom-up from the claims themselves, then each was scored by distinct-firm convergence — the number of separate firms independently voicing it. The five bands (Consensus 26+, Interesting 16–25, Novel 10–15, Sharp 2–9, Lone 1) reflect how crowded a bet is, on the premise that signal is inversely proportional to how many people already hold it. Crowded themes are real but priced-in; sparse themes are where edge, if any, hides.

This is a map of what credible investors are saying, not an endorsement of any claim. Every thesis is attributed to its source below.

📇 Firm Index — who shows up, how often, and where

Every firm in the write-up, alphabetical. Themes = how many distinct convergence themes the firm appears in. Articles = distinct source essays/posts cited for that firm. Sectors touched = the union of sector tags across the themes it voices.

FirmThemesArticlesSectors touched
151727VC model
500Global25AI appsMacroMoatsSoftwareVC model
809084AI appsAI labsDefenseEnergyMacroNatural ResourcesRoboticsServicesSoftware
8VC74DefenseEnergyMacroNatural ResourcesRoboticsServicesSoftware
a16z114AI appsAgentsConsumerDefenseHealthcareMoatsRoboticsSoftware
Abstract22MacroMoatsVC model
Anthology25AI appsMacro
AStar13VC model
BasisSet74AI appsAI labsAgentsMacroMoatsSoftwareVC model
Bessemer46AgentsHealthcareRoboticsSoftware
BOND11AI appsAI labs
BoostVC25EnergyHealthcare
BoxGroup34MacroMoatsVC model
Browder66AgentsConsumerCryptoFintechMacroMoatsVC model
Caffeinated21EnergyMacroMoatsVC model
Contrary47AI appsMacroServicesSoftware
Conviction82AI appsAgentsHealthcareMacroRoboticsServicesSoftwareVC model
Cowboy16Moats
CRV56AI appsAI labsConsumerMoatsSoftwareVC model
DayOne56ConsumerFintechHealthcareMacroRobotics
DCVC113DefenseEnergyHealthcareMacroNatural ResourcesRoboticsServicesSoftware
Emergence66AI appsAI labsAgentsMoatsServicesSoftware
Endeavor14AI appsMacroMoatsSoftware
Eniac46DefenseMacroMoatsNatural ResourcesRobotics
Felicis36ConsumerMoatsSoftware
First Round17Moats
FirstRound27AI appsAI labsMoatsServicesSoftware
Floodgate37AI appsAI labsConsumerMoatsSoftware
Flybridge17Agents
Forerunner33ConsumerMoats
FounderCollective310AgentsMoatsSoftwareVC model
Founders Fund34AI labsDefenseMacroNatural ResourcesSoftware
Fusion Fund41AI appsAI labsEnergyRobotics
General Catalyst125AgentsConsumerCryptoDefenseEnergyFintechHealthcareMacroMoatsNatural ResourcesRoboticsServicesSoftware
Goodwater14AI appsMacroMoatsSoftware
Greylock65AI appsAI labsAgentsConsumerMoatsSoftwareVC model
Haystack57ConsumerMacroMoatsVC model
Headline23AI appsAI labsMoatsServicesSoftware
HustleFund34AI appsMacroMoatsSoftware
IAVentures26ConsumerMoatsVC model
Index69AI appsAI labsAgentsConsumerMoatsServicesSoftware
Initialized812AI appsAI labsEnergyHealthcareMacroMoatsSoftwareVC model
Interplay26AgentsConsumerCryptoFintechServicesSoftware
Khosla107AgentsConsumerEnergyHealthcareMacroNatural ResourcesRoboticsServices
Lightspeed311AgentsConsumerMoats
Maveron76AgentsConsumerMoatsSoftware
Neo13VC model
NfX68AI appsAI labsAgentsConsumerMacroMoatsSoftwareVC model
Notable19Software
Pantera24AI appsFintechMacroMoatsSoftware
Paradigm512AgentsConsumerCryptoFintechMacroSoftware
Pear13MacroVC model
Pebblebed56AI labsAgentsConsumerCryptoFintechHealthcareMoatsSoftwareVC model
Redpoint84AI appsAI labsAgentsConsumerCryptoFintechMacroSoftware
Riot21DefenseRoboticsSoftware
Sequoia67AI appsAI labsAgentsConsumerCryptoFintechMoatsSoftwareVC model
Slow69AI appsAI labsConsumerFintechMacroMoatsSoftwareVC model
Soma12AI labsMoats
Spark47AI labsConsumerMoatsVC model
SPC67AI appsAI labsAgentsConsumerMoatsSoftware
Susa46AI appsAgentsMoatsSoftware
SVAngel23ConsumerMoatsVC model
TenOneTen41AI appsAgentsConsumerCryptoFintechMoatsSoftware
Thiel Capital34AI appsMacro
Thrive310AI labsHealthcareMoatsServicesSoftware
Tuesday12AI appsMacroMoatsSoftware
USV910AgentsConsumerCryptoDefenseFintechMoatsNatural ResourcesSoftwareVC model
WorkBench67AI appsAgentsMoatsSoftware
YC31AI apps

Sources

  1. [157] 1517 — Flux Capacitor 2026: Applications Open for 1517 Angel Check Program ↩
  2. [112] 1517 — Inside v. Outside Context Problems ↩
  3. [90] 500Global — Rewriting the Venture Capital Playbook for the AI Era ↩
  4. [51] 8090 — Armada $131M round announcement — 'The future of AI compute isn't built, it's deployed' ↩
  5. [11] 8090 — Reindustrialization and Deep Tech Investing with 8090 Industries' Grant Brown ↩
  6. [110] 8090 — World's Rich Bet On VC Firm With 275% Gains That Shuns AI (standing signature thesis) ↩
  7. [33] 8VC — The Future of AI, Defense & America — Joe Lonsdale (Global Alts Miami 2026) ↩
  8. [4] 8VC — Top Tech Investors: This AI Wave Is 10X Bigger Than SaaS (8VC podcast, Alex Kolicich & Jack Moshkovich) ↩
  9. [12] a16z — Big Ideas 2026: Part 1 ↩
  10. [1] a16z — Big Ideas 2026: Part 2 ↩
  11. [153] a16z — Big Ideas 2026: Part 3 ↩
  12. [64] a16z — Humans Are for Ideas, AI Is for Execution ↩
  13. [86] Abstract — Ramtin Naimi - Building Abstract (Invest Like the Best EP.435) [interview, non-publisher] ↩
  14. [144] Anthology — 2025 Mid-Year LLM Market Update ↩
  15. [82] Anthology — 2025: The State of Generative AI in the Enterprise ↩
  16. [111] AStar — How AI Is Rewriting Seed Stage Investing (Uncapped Ep. 49 transcript) ↩
  17. [119] BasisSet — Announcing Our $250M Fund IV ↩
  18. [106] BasisSet — Our Favorite AI Workflows of 2025 ↩
  19. [115] BasisSet — Scaling human potential with artificial intelligence (firm thesis page) ↩
  20. [66] BasisSet — You Don't Need to be First ↩
  21. [10] Bessemer — Defense tech roadmap: Five frontiers for 2026 ↩
  22. [14] Bessemer — State of Health AI 2026 ↩
  23. [53] Bessemer — The AI pricing and monetization playbook ↩
  24. [46] Bessemer — The State of AI 2025 ↩
  25. [107] BOND — Trends — Artificial Intelligence ↩
  26. [165] BoostVC — Request for Startups: Longevity/Bio ↩
  27. [166] BoostVC — Why We're Leading Radiant's $300m Series D ↩
  28. [87] BoxGroup — Raising Your Seed Round (David Tisch talk) ↩
  29. [91] Browder — 20VC: Turning Peter Thiel's $100K into a $10M Angel Portfolio + AI's Winners and Losers (Joshua Browder) ↩
  30. [154] Browder — AI4 2025 talk: 5 Key Lessons From Building and Investing in Billion-Dollar AI Companies ↩
  31. [98] Browder — The AI Personal Stack (Newcomer / Cerebral Valley) ↩
  32. [162] Browder — The Consumer Rights Revolution (Joshua Browder, DoNotPay) ↩
  33. [89] Caffeinated — Our Philosophy (firm site) ↩
  34. [81] Contrary — The Case for Specialized AI ↩
  35. [58] Contrary — The Vertical AI Playbook ↩
  36. [5] Conviction — Startup Ideas — The Year of Cheap, Bespoke Code ↩
  37. [138] Conviction — State of Startups and AI 2025 (Sarah Guo, Conviction) ↩
  38. [102] Cowboy — Where we're looking to invest: AI-powered professional services, a $3T opportunity ↩
  39. [132] CRV — Consumer investors say their category isn't dead, it's different (panel, Saar Gur cited) ↩
  40. [84] CRV — Give Money Back (Reid Christian) ↩
  41. [29] CRV — Series A Metrics VCs Expect in 2026 ↩
  42. [74] CRV — Venture has never been more crowded, or more complacent (Reid Christian) ↩
  43. [59] CRV — What Is NRR? Complete Guide to Net Revenue Retention ↩
  44. [73] DayOne — Day One Ventures $150M 'Anti-AI' framing (Fund III, secondary coverage) ↩
  45. [150] DayOne — Day One Ventures secures $150M Fund III: Aims to redefine future of human ↩
  46. [155] DayOne — Masha Bucher 2025 Investment Review ↩
  47. [158] DayOne — Why Day One Ventures' Masha Bucher thinks VCs and storytelling go hand-in-hand ↩
  48. [16] DCVC — 2025 Deep Tech Opportunities Report: An American Industrial Renaissance ↩
  49. [6] DCVC — How the Quantum Race Will Be Won on the Factory Floor ↩
  50. [65] Emergence — Adaptable AI for an Adapting World ↩
  51. [56] Emergence — Should Your SaaS Company Become a Services Business? ↩
  52. [104] Emergence — The Next Token Supercycle: Opportunities in Long-Horizon Inference ↩
  53. [108] Emergence — Why AI-Native Services, and Why Now ↩
  54. [140] Endeavor — Silicon Valley got crowded again. Here are 400 reasons why we're staying Elsewhere. ↩
  55. [50] Eniac — Eniac portfolio signal: Origin Lab / world-model data (firm stories page, 2025) ↩
  56. [3] Eniac — Nihal Mehta investor interview (April 2025) ↩
  57. [76] Felicis — 2025 Annual Letter: An Estimation of Underestimations ↩
  58. [61] Felicis — Q4 '25 Cloud Update (What's going on with software?) ↩
  59. [101] First Round — Inside Artemis' "AI vs AI" War (Josh Kopelman interview) ↩
  60. [28] FirstRound — So You Want to Hire a Forward Deployed Engineer ↩
  61. [75] Floodgate — Era of Mass Cognition ↩
  62. [27] Floodgate — Languaging: The Software of Belief ↩
  63. [167] Floodgate — Why I'm Supporting the Conjecture Institute ↩
  64. [122] Flybridge — The Internet is Closing ↩
  65. [130] Forerunner — Human Insight in the Age of AI (Announcing Fund VII) ↩
  66. [35] Forerunner — Proactive Measures: Be the CEO of Your Own Life ↩
  67. [54] FounderCollective — 6 Takeaways from Collective Future NYC ↩
  68. [117] FounderCollective — Not Every Company Is Built for the Big Leagues ↩
  69. [100] FounderCollective — The Buy Window ↩
  70. [32] Founders Fund — No Solvency, No Security ↩
  71. [163] Founders Fund — What Happened to the Future? (Hereticon II — End of the World's Fair) ↩
  72. [8] Fusion Fund — Next phase for AI value: infrastructure, not models (Lu Zhang interview) ↩
  73. [34] General Catalyst — Built to Endure ↩
  74. [13] General Catalyst — Europe's AI Transformation in Services ↩
  75. [9] General Catalyst — Realizing Resilience ↩
  76. [52] General Catalyst — The Agentic Commerce Opportunity ↩
  77. [57] General Catalyst — The Future of Services ↩
  78. [142] Goodwater — Goodwater consumer AI thesis (insights hub synthesis) ↩
  79. [48] Greylock — Reid Hoffman: Five Predictions About AI in 2026 (AI & I) ↩
  80. [40] Greylock — The Intelligent Future (Reid Hoffman & Saam Motamedi) ↩
  81. [19] Greylock — The New New Moats ↩
  82. [135] Haystack — Lightspeed Generative Now: Venture Capital Trends in the Age of AI (with Michael Mignano) ↩
  83. [85] Haystack — New Funds for Haystack ↩
  84. [72] Haystack — The New Rules of Early-Stage Venture in the AI Era (with Semil Shah) ↩
  85. [31] Headline — Introducing the Top 50 Vertical SaaS Companies to Watch in 2025 ↩
  86. [109] Headline — The Infra Stack Reset: 10 Trends Reshaping Spend in 2025 ↩
  87. [83] HustleFund — The wide path: Why most current valuation dynamics are not here to stay (and what will) ↩
  88. [71] IAVentures — Inside Roger Ehrenberg's new firm (INTERVIEW/profile) ↩
  89. [88] IAVentures — Roger Ehrenberg interview: Why VC Returns Will Get Worse (INTERVIEW, not firm essay) ↩
  90. [43] Index — Acts of Claude: Building with Agents, a 'Work in Progress' ↩
  91. [24] Index — Erin Price-Wright thesis — human-shaped vs computer-shaped tasks ↩
  92. [68] Index — From $300B to $10 Trillion: How Moment Became the AI Operating System for Investment Management ↩
  93. [131] Index — Index Perspectives — Infrastructure & AI positioning ('inference is the new runtime') ↩
  94. [63] Index — The Execution Gap No Model Can Close (Wonderful, Index-backed) ↩
  95. [17] Initialized — Agents at Work ↩
  96. [67] Initialized — Backing Garage: Transforming How America's Critical Equipment Is Bought and Sold ↩
  97. [105] Initialized — Initialized Backs Valar Atomics to Build Nuclear Gigasites ↩
  98. [118] Initialized — Initialized Leads $7M Round for Enhanced Radar ↩
  99. [134] Initialized — The Next Great Food Infrastructure Company Won't Be Built Around a Single Ingredient ↩
  100. [97] Interplay — The Future of Commerce: Agents are Rebuilding B2B Procurement from the Ground Up ↩
  101. [38] Khosla — A Roadmap to AI Utopia (Khosla Ventures / TIME) ↩
  102. [156] Khosla — Vinod Khosla AI job prediction interview (Fortune) ↩
  103. [2] Khosla — Vinod Khosla on the End of Jobs and the Future of Capitalism (Newcomer) ↩
  104. [125] Khosla — Vinod Khosla: today's 5-year-olds won't need a job (Business Insider) ↩
  105. [69] Lightspeed — The Realness of Live Sports in an AI Generated World ↩
  106. [123] Lightspeed — The Remarkably Rapid Rollout of Foundational AI Models at the Enterprise Level: A Survey ↩
  107. [127] Lightspeed — Tokens Aren't Fungible ↩
  108. [36] Maveron — AI Is Transforming CPG Distribution ↩
  109. [129] Maveron — The Invisible Rebuild: How AI Is Quietly Reshaping the Backbone of CPG ↩
  110. [62] Maveron — The Next Cultural Powerhouse: Why Gen Alpha Can't Be Ignored ↩
  111. [126] Maveron — What Consumers Are Doing Now And What Comes Next (Consumer 2026 report) ↩
  112. [114] Neo — Y Combinator Rival Neo Is Harder To Get Into Than Harvard ↩
  113. [79] NfX — 3 Waves of Successful Generative AI Startups ↩
  114. [137] NfX — AI Zen ↩
  115. [30] NfX — Generative AI Market Map and 5-Layer Tech Stack ↩
  116. [47] NfX — How 'Venture Capital 3.0' Impacts Founders in the AI Age ↩
  117. [116] NfX — The 3-Person Unicorn Startup ↩
  118. [60] Notable — The Future of Voice AI: How Standalone Companies Can Thrive in a Speech-to-Speech World ↩
  119. [143] Pantera — Forward Outlook on the Opportunity Set for Fund V ↩
  120. [149] Pantera — Market Outlook for 2026 + Nine Predictions for 2026 ↩
  121. [49] Paradigm — Introducing Paradigm Predictions ↩
  122. [94] Paradigm — Introducing the 2026 Paradigm Fellowship ↩
  123. [168] Paradigm — Opportunity Markets ↩
  124. [136] Pear — Pear's Request for Startups — Plan mode for every knowledge worker (Shravan Reddy) ↩
  125. [164] Pebblebed — "Intelligence" is not a technical term. Hardness is. ↩
  126. [133] Pebblebed — Software is dead. Long live software ↩
  127. [95] Pebblebed — The End of Per-seat SaaS ↩
  128. [151] Pebblebed — The Internet Will Have More Agents Than Humans ↩
  129. [96] Redpoint — 25 for '25: The App Renaissance Arrives ↩
  130. [44] Redpoint — AI Update: 2025 InfraRed Report ↩
  131. [120] Redpoint — From Middlemen to Machines: The New Economic Order ↩
  132. [41] Redpoint — From Pilots to Platforms: Picks for '26 ↩
  133. [7] Riot — From Generalist to Deep Tech Specialist: A Conversation with Will Coffield (Stifel Forge Series) ↩
  134. [42] Sequoia — 2026: This is AGI ↩
  135. [18] Sequoia — AI in 2026: The Tale of Two AIs ↩
  136. [99] Sequoia — AI Retail Opportunity ↩
  137. [25] Slow — 6 Predictions For 2026 ↩
  138. [160] Slow — Investing Directly in People Is the Future of VC ↩
  139. [113] Slow — Slow Conviction (Should I vs Can I) ↩
  140. [22] Soma — Soma Capital homepage (mission / portfolio framing) ↩
  141. [128] Spark — Personality, humanism, Poke, and raising the ceiling (Hallway Chat) ↩
  142. [77] Spark — Spark Capital VC Nabeel Hyatt sees robotics reshaping the gig economy ↩
  143. [21] Spark — Yasmin Razavi turned a $75M check into a $3B AI windfall (Forbes Midas) ↩
  144. [26] SPC — Avoiding The Eye of Sauron ↩
  145. [39] SPC — Entering the Post-Prompting World ↩
  146. [20] SPC — The Silicon Valley Playbook is Broken (with Elad Gil) ↩
  147. [148] Susa — Compounding Competitive Advantage, Part II: Scale Economies Shared ↩
  148. [103] Susa — Deep Tech Companies Are Built Different ↩
  149. [145] Susa — The AI Blind Spot That Gets Bigger as AI Takes On More ↩
  150. [152] Susa — The AI Moment, Part IV: Lightning Round ↩
  151. [78] SVAngel — AI Open Letter — Build AI for a Better Future ↩
  152. [92] TenOneTen — Money at Machine Speed ↩
  153. [80] Thiel Capital — A Mind-Bending Conversation with Peter Thiel (NYT 'Interesting Times' / 'A.I., Mars and Immortality') ↩
  154. [15] Thrive — Invest Like the Best (Josh Kushner): Concentration and Conviction / Thrive in a New World ↩
  155. [23] Thrive — Invest Like the Best (Josh Kushner): Disruption from the inside out ↩
  156. [139] Thrive — TIME100 AI 2025 — Joshua Kushner / Thrive Holdings thesis ↩
  157. [141] Tuesday — Tuesday Capital Adds Ethan Imboden as Partner (press release) ↩
  158. [161] USV — 10 Highlights from USV's 2025 Annual Meeting ↩
  159. [159] USV — Electroflow ↩
  160. [70] USV — Glif ↩
  161. [93] USV — History Rhymes? ↩
  162. [124] USV — Our Investment in Anon ↩
  163. [37] USV — Suno ↩
  164. [45] WorkBench — [Work-Bench Research] Exploring the Future Compute Stack for AI Agents ↩
  165. [55] WorkBench — Enterprise Weekly #538: The New Economics of AI Applications ↩
  166. [147] WorkBench — Enterprise Weekly #543: What We're Excited About Going into 2026 ↩
  167. [146] WorkBench — The Rise of the Agent Runtime ↩
  168. [121] YC — RFS Summer 2026: AI-Powered Agriculture ↩

Generated by Galileo 🔭 · June 5, 2026