I have spent time on both sides of how markets interpret technological change. Earlier  in my career, working across public markets and multiple asset classes, the lens was  largely one of risk management, allocation frameworks, and benchmark construction.  Early-stage venture capital offers a different vantage point. You sit closer to where  companies are being built before they are priced, before they are widely known, and  often before the sectors they may eventually reshape have fully recognised the pressure coming their way. 

From this side of the market, the more important issue increasingly looks informational  rather than return driven. 

The market now appears to operate across two parallel layers. One is the disruption  layer: the systemic forces actively revising the assumptions on which a significant  portion of institutional capital is priced. The other is the innovation layer: the  companies being built on top of those forces, mostly private, mostly early-stage, and  often outside the institutional capital stack entirely. 

Most large portfolios already have meaningful exposure to the first layer and limited  exposure to the second. The reasons are understandable: mandate restrictions,  governance frameworks designed for later-stage evaluation, and fee structures difficult  to justify against public market benchmarks. Investors best positioned to absorb long duration illiquidity are often the furthest from the layer where the earliest forward  signals emerge. 

To test whether the observation held up beyond intuition, I built a framework and  applied it to publicly disclosed holdings across some of the largest pension funds in  developed markets using December 2025 positions.

I am working from public disclosure and not claiming to have a full picture of any fund’s actual  positioning. PwC’s 2025 Global Investor Survey found that 53% of investment professionals  describe the companies they invest in as having high or extreme exposure to technological  disruption.  

The framework 

The method maps every material position against two axes. 

Disruption Risk measures the degree to which a position depends on business model premises  currently being revised by systemic forces. Systemic and cyclical pressures behave differently. Cyclical pressure recovers when conditions normalise; systemic disruption involves a  permanent revision to the underlying logic that supported a valuation: the demand  assumptions, the competitive barriers, the regulatory framing. Positions in the second category  require an explicit view on the pace of that revision, not just its direction. 

Innovation Exposure measures the degree to which the portfolio participates in the companies  and capital generating the disruption, rather than merely receiving it. This axis has three layers  and conflating them is the most common source of false confidence in how institutional  portfolios describe their innovation positioning. 

– The infrastructure layer covers companies building the computational, physical, and financial  substrate on which the next economy operates: hyperscalers, chip manufacturers,  semiconductor tooling, and energy infrastructure supplying the capacity for AI workloads.  Owning these is a bet on the buildout of a platform, not on what gets built on top of it. 

– The application layer covers companies actively deploying that infrastructure to displace  incumbent business models, among them AI-native financial services providers compressing  traditional banking margins, climate technology companies repricing the energy sector’s asset  base, and health technology businesses restructuring diagnostics and distribution in ways that  revise chronic disease device economics. Most significant players in this layer are private at the  stage where the displacement economics are most asymmetric. 

– The formation stage covers pre-revenue or early-revenue companies at the point where their  eventual displacement of incumbents is speculative, but where the return profile is most  asymmetric, and where the earliest signals about which categories are forming are clearest.  Most institutional portfolios have very limited exposure to this layer, for reasons I think are  understandable: mandate restrictions that explicitly prohibit pre-revenue investing, fee  structures that may be difficult to justify against public market benchmarks, governance  frameworks not designed to evaluate companies without revenues, and the simple fact that  top-quartile managers at this stage are capacity-constrained.  

Within the Disruption Risk axis, positions are classified against five vectors. Each represents a  distinct mechanism by which business model premises are being revised.

– AI Displacement: business models dependent on labour-intensive cognitive functions that AI  agents, automation, and decision engines are demonstrably replacing. Retail banking  operations, insurance processing, legal and professional services, employment intermediaries.  Positions here require a view on margin trajectory over a three-to-seven-year horizon that  standard sector analysis is not calibrated to provide. 

– Climate Transition: operating economics priced on carbon-intensive premises that policy,  technology, and shifting capital costs are actively revising. Fossil fuel extraction, coal dependent infrastructure, conventional automotive, carbon-intensive materials. The vector is  asymmetric: risk for those holding premises being revised, and significant opportunity for  those positioned on the other side. 

– Biotech and Health Disruption: disease burden, utilisation, and drug development premises  being revised faster than most models anticipated. AI-accelerated drug discovery is  compressing the timeline from target identification to clinical candidate. AI diagnostic tools are  restructuring clinical workflows and the procedural volumes that underpin medical device and  hospital economics. Both are moving faster than current equity valuations in the sector tend to  reflect. 

– Regulatory Fragmentation: global business models built on cross-border data flows, payment  rails, and platform economics that assumed regulatory convergence, now facing divergence  through data localisation requirements, AI Act compliance variation, sanctions regimes, and  trade policy bifurcation. 

– Geopolitical Supply Chain: production and distribution architectures exposed to fracture lines in  semiconductor trade restrictions, critical mineral dependencies, energy supply concentration,  tariff regimes, and conflict proximity. This vector cuts across sectors in ways that geography based analysis tends to miss. 

The five vectors do not operate on the same timeline. Geopolitical Supply Chain risk is active in  certain positions today. AI Displacement is revising margins across most affected sub-sectors on  a medium-term horizon. Climate Transition plays out across decades but carries non-linear  policy shock risk that can accelerate repricing abruptly. Understanding which vector is relevant  to a given position, and at what velocity, matters as much as identifying it. 

Innovation exposure also has a geography. The disruption vectors hitting most large  institutional portfolios are global in origin: AI compression of banking margins, energy  transition pressure on resource companies, pharmaceutical innovation revising device market  economics. The companies being built on the other side of those trades are distributed across  San Francisco, Stockholm, London, Singapore, and increasingly across markets in APAC, MENAP,  and Continental Europe. Formation-stage capital that is geographically domestic to a given  fund’s home market may be excellent capital, but it is not necessarily positioned on the same  displacement vectors the fund is most concentrated in.

What I found 

Applied to publicly disclosed holdings across the cohort, the framework surfaces four  observations consistent enough across funds, jurisdictions, and portfolio structures to warrant  examination. All are drawn from the visible portion of each portfolio only. 

AI infrastructure concentration is coordinated, not diversified 

Across the funds looked at, AI infrastructure concentration tends to cluster between 15 and  25% of the total equity book, concentrated in the same eight to ten names: the dominant  hyperscalers, the leading semiconductor manufacturer, the principal chip designer driving the AI compute cycle, the key tooling companies in the semiconductor supply chain. When these  names appear at consistent weights across multiple institutional equity books, correlation  analysis still classifies them as diversified across sectors. They are not. They represent  coordinated exposure to one resolution of the AI transition: that value accrues primarily to  infrastructure incumbents, that the application layer does not materially compress their  margins, and that current capital expenditure cycles are well-calibrated to actual demand. 

The leading chip designer appeared as one of the largest single equity positions, at weights of  between 3 and 6% of the total equity book. The position is real conviction, and the returns have  validated it. The question the framework is designed to ask is a different one: whether owning  the infrastructure layer, at valuations that already price in significant AI optionality, constitutes  genuine innovation exposure or consensus market exposure to the current cycle’s winners.  These are concentrated bets on one resolution of the AI transition. 

A further observation on the semiconductor positions specifically: the largest chip manufacturer  in most portfolios spans two disruption vectors simultaneously, functioning as a near-pure  infrastructure-layer bet and one of the highest concentrations of geopolitical supply chain risk  in the same book. Most advanced AI chips driving the current cycle are manufactured in a single  jurisdiction. Standard sector analysis does not surface both dimensions together, and most risk  reporting does not either. 

Banking sector exposure is roughly double what equity reporting shows 

The banking sector exposure visible in the equity book consistently understated the actual  position once fixed income holdings in those institutions were aggregated. The largest domestic  and global banks appear as meaningful equity positions; they then reappear across the fixed  income book in bonds at comparable scale. One name, the same AI Displacement vector, in a  separate part of the portfolio report. Asset-class-level reporting is designed to show  diversification across buckets, not concentration within a single disruption vector across them. 

The practical implication: an investment committee reviewing banking sector disruption risk  from equity-only analysis is looking at roughly half the actual exposure. The fixed income 

component compounds the equity concentration rather than offsetting it and adds credit and  counterparty dimensions that sector analysis is not built to capture. For funds where domestic  banks are simultaneously the largest equity positions, significant fixed income issuers, and  primary cash counterparties, the cross-asset aggregation produces a number that is materially  larger than any single-asset-class view would suggest. 

Drug innovation and device exposure are reported as separate positions in one trade 

Across the cohort, funds hold positions in the leading GLP-1 drug manufacturers alongside  meaningful positions in the global leaders in chronic disease device categories whose volumes  are being revised by those same drugs. The two positions sit in different sector classifications, in  some cases in different asset classes, and are reported in ways that make them appear as  unrelated exposures. A disruption lens reads them as a single trade: a long position in a  pharmaceutical innovation cycle and a long position in the device market assumptions that  cycle is actively revising. 

The GLP-1 dynamic is the most visible current instance of this pattern, but the underlying  mechanism is not specific to it. AI-accelerated drug discovery compressing the timelines to  clinical candidates across multiple therapeutic categories, diagnostic tools restructuring the  procedural volumes that underpin medical device economics, pharmaceutical innovation  revising the chronic disease burden assumptions on which insurance and device businesses are  priced: the same cross-sector revision dynamic recurs across the health book in ways that  sector-level reporting does not aggregate. The pattern is not a risk management failure;  standard classification frameworks were not built to show it. 

Formation-stage exposure is limited or absent across most portfolios 

Across the funds I examined, the pattern holds regardless of fund size, total alternatives  allocation, or the sophistication of the private markets programme: exposure to early-stage  venture capital is either absent from the disclosure or limited in scope and usefulness. The  private and alternative books are professionally managed and, in several cases, distinguished.  The gap is in access to the innovation signal, not in the quality of what surrounds it. 

The geographic pattern compounds this. Where early-stage venture exposure exists, it tends to  be domestic, concentrated in a small number of manager relationships, and calibrated more to  benchmarking logic than to the geography of disruption. The displacement vectors hitting the  largest holdings in most portfolios are not playing out domestically first. AI compression of  banking margins, energy transition pressure on resources companies, software-led  transformation of enterprise services: these are being built in San Francisco, Stockholm,  London, and Singapore. Domestic early-stage capital hedges against domestic displacement risk.  For portfolios whose largest disruption concentrations are global, the geographic mismatch is  material.

The categories forming at pre-seed and seed are the most direct leading indicator of where  competitive pressure on existing holdings is building. Public market pricing captures that signal  after it has moved. The signal appears upstream first. 

Three questions sit underneath the framework 

These are observations from public disclosure, not a full audit of any fund’s positioning. Three  questions keep coming back to me when I look at portfolios through this lens. 

1. The calibration question. Where in the portfolio is there a long position in business model  premises currently being revised by systemic forces, and is that a conscious, priced-in view or  an inherited position that has not been reassessed against the velocity of the revision it faces?  Not sectors in broad terms, but specific premises: that branch-based banking retains pricing  power through an AI-enabled challenger cycle; that chronic disease device volumes are durable  against pharmaceutical innovation revising their end markets; that semiconductor  infrastructure at current valuations represents innovation access rather than consensus market  exposure; that global platform economics persist as regulatory convergence fragments.  Investment committees tend to have views on which sectors face disruption. They less often  have documented positions on the speed. 

2. The position question. Across the total portfolio including private markets, how much  capital sits at the earliest stage of the displacement vectors the portfolio is most concentrated in,  and is it geographically matched to where those displacements are being built? For most  portfolios the honest answer reveals a gap that qualitative investment discussions tend not to  expose. The early-stage exposure that exists is typically small, predominantly domestic, and  calibrated more to benchmarking logic than to displacement geography. 

3. The horizon question. For each disruption vector that is material in the portfolio, what is  the fund’s view on the pace of repricing, and does the current position reflect that view or is it  inherited from a prior allocation decision that assumed a different velocity? A position  defensible under a seven-year transition horizon carries different risk if that transition is  compressing to three. The pace question is the one most consistently absent from investment  committee discussions about disruption. 

The price discovery gap 

The most valuable thing early-stage investing offers large institutional portfolios looks  informational rather than return-driven. Across full-year 2025, 136,000 founders applied to  build with Antler globally. We backed 394 of them, roughly one in 346, approximately twelve  times more selective than Harvard. What those founders choose to work on is often a leading  indicator of where competitive pressure is building long before it becomes consensus.

Consumer AI dominates the headlines. Yet consumer technology's share of Antler's annual  investments fell from 27% in 2022 to 15% in 2025. The venture layer signal has been moving  elsewhere: toward enterprise infrastructure, applied AI at the sector level, physical AI, and deep  technology categories that remain largely absent from later-stage deal flow. The number of  physical AI companies in Antler's annual cohorts was zero in 2022, six in 2023, fifteen in 2024,  and twenty-five in 2025. That is a category forming in real time, twelve to eighteen months  before it begins appearing meaningfully in growth-stage valuations. Portfolios with no visibility  into this layer will likely encounter it only once it starts repricing the industrial and hardware  positions they already hold. 

Quantum computing, physical AI, applied science at seed: these are not abstract future themes.  They are where capable founders are choosing to build across twenty-six markets, filtered  through one of the most selective founder funnels in early-stage investing. 

The deeper irony is the mismatch for long-duration investors. Pension funds, sovereign wealth  funds, and endowments are structurally among the best positioned to absorb illiquidity, hold  through J-curves, and participate in long-horizon technological shifts. Yet many portfolios  remain concentrated in assets where information about disruption arrives only once repricing  has already begun. 

The pattern recurred across the portfolios analysed, driven less by idiosyncratic investment  decisions than by institutional incentive structures that make tracking-error minimisation  easier to maintain than disruption positioning. The gap is visible from public disclosure alone,  which ultimately prompted the development of a simplified self-assessment version of the  framework. The more important question it raises is not whether every institutional portfolio  should allocate to early-stage venture capital. It is whether portfolios exposed to systemic  technological change can afford to remain structurally distant from the layer where the earliest  signals about that change first emerge.

^{A B O U T A N T L E R I N T E L L I G E N C E} 

^{Antler is one of the most active early-stage venture firms globally, with US$1.3B+ AUM, operations across 26 markets,  and a portfolio of 1,900+ companies spanning every major technology market. We invest from pre-seed and seed,  where much of our portfolio sits, through to Series C via regional early-stage funds and global follow-on vehicles. That  structure gives us something few investors have: a longitudinal, cross-geography dataset of founder behaviour,  category formation, and technology adoption that begins before most institutional investors are in the room.} 

^{Antler Intelligence is our research and market insight function designed for an LP audience. The views expressed  here are based on publicly available disclosure documents and Antler’s proprietary analytical model. They represent  a research perspective on portfolio construction dynamics, not investment recommendations or the position of any  Antler fund. We publish this analysis because the early-stage vantage point is underrepresented in institutional  market discourse, and because our LPs and ecosystem partners benefit from understanding what we are seeing  before the market prices it in.} 

^{D A T A A N D D I S C L O S U R E} 

^{This article may contain forward-looking statements based on current assumptions, which are subject to change. No obligation is  undertaken to update this material. It is not intended for distribution in any jurisdiction where such distribution would be contrary to  applicable law.} 

^{Holdings data referenced in this analysis is sourced from publicly available position-level portfolio disclosures published by institutional  funds in accordance with their regulatory requirements. Data reflects December 2025 portfolio positions. Funds were selected based on disclosure quality and scale; no fund is named in this analysis. All observations are stated as cohort-level patterns, not attributions to  specific institutions. Externally managed mandates hold underlying securities not visible in direct disclosure; actual exposure to  individual sectors, issuers, and asset classes will differ from disclosed positions. This analysis should not be taken as a complete  characterisation of any fund’s investment approach, mandate, or performance.} 

^{The disruption vector model applied here is a qualitative analytical lens developed by Antler, not a quantitative risk model. Vector  classifications reflect Antler’s assessment of the systemic forces affecting relevant business models at time of writing; reasonable  analysts may reach different classifications.} 

^{Antler operates in the formation-stage market that this analysis identifies as underrepresented in most institutional portfolios. Readers  should weigh the analysis with that commercial interest in mind. This document does not constitute investment advice, a personal  recommendation, or an offer to invest in any Antler fund or vehicle.} 

^{F O R I N V E S T M E N T T E A M S I N T E R E S T E D I N A P P L Y I N G T H I S  A P P R O A C H} 

^{A self-assessment diagnostic that produces an indicative read of innovation access and disruption exposure at  portfolio level is available for illustrative purposes. Twenty questions, approximately ten minutes. For a conversation,  contact intelligence@antler.co.}

‍