Inner Loop Intelligence

Graph View

17 min read

AI Disruption — Knowledge Economy Destruction & SaaS Moat Erosion

AI will decimate middle management and entry-level knowledge worker jobs in the next few years, with college graduates unable to find employment.

Thesis Health

No signal data available yet.

Thesis / Overview

AI will decimate middle management and entry-level knowledge worker jobs in the next few years, with college graduates unable to find employment.

Key claims

🟢 Supporting    🔴 Challenging    🟡 Contested    💬 Commentary

🟢 Supporting Evidence (39)

  • Market sentiment rotating from offense (knowledge/AI) to defense (tangible assets). Three scenarios: software sell-off is temporary panic, AI capex continues regardless, or rotate into HALO (Heavy Assets, Low Obsolescence) - energy, materials, staples. [@Gaetan Warzee] (2026-04-07 — fresh) → source
  • Thinking about non-consensus places AI might impact first. BPO/SI space has come up as worth tracking as a canary and possible short opportunity. [@Will B] (2026-04-07 — fresh) → source
  • Morgan Stanley maps AI disruption risks across leveraged credit. BDCs most exposed at ~26%, private credit CLOs ~19%, US loans ~16%. 50% rated B- or lower, 80% private issuers. Software loans down ~3% YTD with 20% trading distressed below $80. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • If knowledge is suddenly ubiquitous and free thanks to AI, investors could fall back on things they can touch - copper mines, oil rigs, refineries, turbine manufacturers. AI seems to be destroying US outperformance rather than solidifying it. [@Gaetan Warzee] (2026-04-07 — fresh) → source
  • Goldman recommends going long select incumbents as ‘fast followers’ with strong domain moats and clear AI roadmaps: Microsoft, ServiceNow, Salesforce, HubSpot, Snowflake, Cloudflare, Palo Alto Networks, CrowdStrike. Tilt gradually from capital-light tech toward capital-heavy AI-insulated sectors like telecoms, industrials, utilities. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • Visser’s ‘multiple compression’ argument: it’s not necessarily that a company won’t exist in 5 years, but should it trade on a 20-30 year multiple when its business may be completely different because of AI? [@James S] (2026-04-07 — fresh) → source
  • If Visser is right, the only assets with genuine long-duration residual claims are those whose value doesn’t depend on business models: gold, land, copper, oil reserves, water rights, timber. Hard assets become the new equity. [@Will B] (2026-04-07 — fresh) → source
  • Today AI eats software, tomorrow we create humanoid robots that eat everything else. It’s not a bad thing, we just created something that is more efficient than ourselves! [@James S] (2026-04-07 — fresh) → source
  • Travis Kling’s framework: AI creates deflation, which is the ultimate boogeyman for debt-laden systems. Fed and government will be forced into accommodation to keep prices flat. This leads to massive wealth inequality and forces UBI. [@Will B] (2026-04-07 — fresh) → source
  • We’re restricting immigration to lower competition for manual labor while ramping up AI to increase competition for knowledge work. In short term margins expand, in medium-to-longer term increased competition forces margins back down. [@Jesse] (2026-04-07 — fresh) → source
  • Gromen warns AI deflation is hollowing out white collar jobs at AI speed, which will collapse corporate profits via consumer credit crisis—unlike the blue collar hollowing which had a long lead-time cushioned by subprime lending. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • AI may be disrupting the very companies that benefited most from the ‘knowledge economy.’ Software and information-based business models have seen major drawdowns. If AI commoditizes knowledge, US equities lose a key pillar of their valuation premium. [@Gaetan Warzee] (2026-04-07 — fresh) → source
  • Jordi is very worried about software. If IGV goes lower, hedge funds could face forced liquidations. Software forward PE is now down from 35 to 19, which is approximately when iPhone was released. [@James S] (2026-04-07 — fresh) → source
  • Become absolute power user - using AI a lot for research and trade idea generation. Allows covering much more ground and found incredible winners from AI research. Has properly engineered a prompt pipeline. [@Antonio Furtado] (2026-04-07 — fresh) → source
  • Pair trade concept: Long agentic AI economy, Short sharing economy. [@Mark Tetreault] (2026-04-07 — fresh) → source
  • Posted article that may explain what’s been weighing on bank stocks recently (referencing private credit concerns). [@thibault] (2026-04-07 — fresh) → source
  • Jordi Visser becoming increasingly worried about private credit. Over past 3-12 months said it was fine, but last 2 videos show serious signs of distress. Overall quite negative about markets. [@James S] (2026-04-07 — fresh) → source
  • Private credit stress being discussed by CDO builders from 2005-08 era. Feels like 2007 to them - asset class driven by AUM/fee growth not risk-adjusted returns. Blue Owl has ‘spiral potential.’ 20-35% discounts appearing in tender offers. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • Peak liquidity losers including private credit are groping for floor after Q1 VaR shock. If private credit can catch bid on peak yields/steeper curve, soft landing; if not, hard landing coming. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • Many products won’t ‘die’ but will lose pricing power, expansion capacity, strategic independence over 12-24 months. Deepest mispricing is confusing features with durable power in agentic world. [@Nicky Adam] (2026-04-07 — fresh) → source
  • AI’s disruptive potential is becoming really disruptive to investment decision-making process — couldn’t sleep thinking about AI impact on private credit markets. [@Nicky Adam] (2026-04-07 — fresh) → source
  • Alex Wang’s daily substack shows AI impact and velocity is massively underappreciated. His posts are like ‘peeking into the future’ while 99.999% of people dismiss it as chatbots taking over. [@Will B] (2026-04-07 — fresh) → source
  • When I talk about next few years when middle mgmt jobs get decimated, and their college graduating kids move back home since they can’t get jobs, I get ‘I never thought of that’ response from very successful individuals on ship. Not pushback, just shocked realization. Majority don’t understand implications. [@Scott Leavitt] (2026-04-07 — fresh) → source
  • Long Hyperscalers CDS to hedge against risks from excessive AI-related capex and potential tech-sector balance-sheet strain. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • A company with less of a moat trades at lower multiple than one with strong moat. AI doesn’t need to bankrupt companies to completely change competitive landscape. Investors shoot first, ask questions later. [@Antonio Furtado] (2026-04-07 — fresh) → source
  • Professional services see productivity gains but clients doing same AI acceleration won’t pay same fees. Healthcare AI productivity increase offset by loss of inbound business as informed patients opt out of visits. [@Michael Moshiri] (2026-04-07 — fresh) → source
  • Early adopters reduce overheads and profit while prices stay at old levels. Then disruptors enter at new marginal cost, early adopters adjust margins back to pre-adoption levels, and non-adopters go out of business. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • Report notes Claude Code allows engineers to run multiple instances working on different parts simultaneously. They become project managers working with compute to create code at 18x speed. 4% of all GitHub code is now AI-generated. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • AI tsunami will be enormously deflationary. What we see in credit and white-collar employment is leading edge of wave displacing lawyers, analysts, accountants, consultants, software developers - high-income, high-spending people. When their purchasing power goes, demand crater is significant. [@Will B] (2026-04-07 — fresh) → source
  • Blue Owl Capital failed to secure ~$4B in debt financing for a major CoreWeave data center project. Lenders cited perceived risks in the overheated AI infrastructure boom, contributing to selling pressure. [@Antonio Furtado] (2026-04-07 — fresh) → source
  • This represents an equilibrium shift in supply/demand curves. Cell phones replaced encyclopedias and phone books; AI will replace some service providers. Not looking good for call centers in India. [@Mark Tetreault] (2026-04-07 — fresh) → source
  • AI trade moving from chips to what has a moat that will never be disrupted. Will Walmart still be here in 30 years? Yes. Framed as SaaS vs Walmart multiples. [@thibault] (2026-04-07 — fresh) → source
  • Deutsche Bank: Model providers unlikely to displace software incumbents, positioning instead as orchestration layer on top of existing systems. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • Sparkline argues AI boom shifting from buildout to adoption phase. AI infrastructure at 2x market valuation, while early adopters in financials/healthcare/industrials trade at valuations similar to AI laggards with 3-5x more AI assets per dollar. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • From Gavin Baker chart - AI is having material impact on productivity. What matters is knowing what questions to ask - that’s ultimately a creative task. Intelligence democratisation doesn’t mean everyone uses it effectively, similar to last 20 years of internet. [@Will B] (2026-04-07 — fresh) → source
  • Citrini Research speculative 2028 scenario shows AI triggering 38% S&P drop and 10%+ unemployment - AI automates tasks not entire jobs, triggering violent restructuring. [@thibault] (2026-04-07 — fresh) → source
  • Ironshield report notes Claude Code allows engineers to become project managers working with compute to create code at 18x speed. 4% of all GitHub code is now AI-generated - replication risk is structural not hypothetical. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • As someone with long market experience, thinking about how tools reshape early-career talent development. Learning came from the grind - building models, refining assumptions. If AI compresses that to minutes, the path to gaining experiential foundation changes dramatically. [@Mike Arnold] (2026-04-07 — fresh) → source
  • Claude for Excel built a 4-sheet valuation model in 20 minutes - read earnings reports, presented assumptions, made adjustments. This means lots of analysts are out of a job. [@James S] (2026-04-07 — fresh) → source

🔴 Challenging Evidence (14)

  • If you have medium-long time horizon and strong conviction we’re in very early innings of AI, then buy while others are fearful. That’s how FU money is made. [@Scott Leavitt] (2026-04-07 — fresh) → source
  • The singularity is overhyped. The spirit of it will take much longer to play out because of immense unappreciated lags in real world manufacturing and changing habits—like global warming, the difference between how it was advertised vs how it is playing out. [@Jesse] (2026-04-07 — fresh) → source
  • At luxury hotels, I want humans around to shape my experience, not robots. Hospitality has an inherently human core—the warmth, the human-to-human connection, which can never come from a robot. [@Nicky Adam] (2026-04-07 — fresh) → source
  • Studying the past drives me to the same opinion. There will be disintermediated workers but eventually new jobs appear. Some people adjust, others withdraw, but everyone finds their new place. [@Mark Tetreault] (2026-04-07 — fresh) → source
  • AI may create more jobs rather than fewer, especially in service industry. More efficient systems lead to increased profits, which could be spent on hiring more employees for better customer satisfaction and brand loyalty. [@Gary Winters] (2026-04-07 — fresh) → source
  • My first instinct is it’ll be a slow burn. You wouldn’t throw away a working system that’s critical to your business on a whim. Transitioning from one system to another with migration and switchover is one of the hardest things to do operationally. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • Companies will operate parallel systems until reliability is proven and the C-Suite gradually dies off. Nothing ever said it was all or nothing. [@Mark Tetreault] (2026-04-07 — fresh) → source
  • The SAAS apocalypse is overdone. AI coding adoption is not moving as fast as the market has priced. Clients have spent a lot on current systems and value comfort, convenience, and relationships. SAAS companies are already off 25-30% and may gain upward momentum. [@Gary Winters] (2026-04-07 — fresh) → source
  • …and how long will those roles last… [@Scott Leavitt] (2026-04-07 — fresh) → source
  • The ‘Citrini doom’ bear case, where rapid AI automation triggers mass job losses and systemic crises in private credit, is wildly overstated. AI disruption will not cause a recession on its own. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • Data centers only part of business for companies like GOOG. They can directly monetize capacity for own purposes. Would ‘buy when others are fearful’. Differentiates from companies solely building raw capacity to sell. [@Scott Leavitt] (2026-04-07 — fresh) → source
  • SaaS doomsday scenario has merit in some cases but is overblown. Companies with multiple business lines won’t go out of business due to AI infrastructure debt. Opportunity for dividend yields. [@Mark Tetreault] (2026-04-07 — fresh) → source
  • The productivity focus is short-sighted. The real opportunity is creating totally new products/services and revamping business models, not just cost savings — similar to internet evolution from eyeballs to Uber and Amazon. [@Scott Leavitt] (2026-04-07 — fresh) → source
  • Disagrees with Jordi Visser’s view that large enterprises are structurally limited and unable to adopt AI. [@Mark Tetreault] (2026-04-07 — fresh) → source

🟡 Contested / Debated (1)

  • Middle management thing is tricky because on one hand they get automated away from today’s standpoint, but on the other hand that’s the only job left for humans - middle managing AI bots going forward. [@Jesse] (2026-04-07 — fresh) → source

💬 Commentary & Context (14)

  • Still need to pick the winners. There will be a lot of losers along the way. [@Gaetan Warzee] (2026-04-07 — fresh) → source
  • Interesting that mood has changed enough for deep value guys to look at software. Sentiment shift in last few months is significant. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • The nuance is not that any rando can vibe code their own software - competition between software studios gets intensified and drives prices down. Like hardware where it’s not about design but manufacturing, it’s far more complicated to setup reliable continuous improvement and deployment pipeline than just design something ‘that works’. [@Jesse] (2026-04-07 — fresh) → source
  • CEOs will need to balance cost cutting and taking risk on trusting AI. Reputation risk probably trumps cost cutting. In 2017 when I left the bank I still didn’t have email access outside physical workstation - all for security and privacy. See more software companies getting squeezed for margins by clients but making it up using AI to cut their costs. [@Gaetan Warzee] (2026-04-07 — fresh) → source
  • The entire system is built around a social contract of education = future deferred income. There won’t be a more militant class than PhDs/MSCs working in hospitality for minimum wage. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • The AI wave will benefit people who are younger, not afraid of code and have a management mindset. It will cause a lot of older people to just retire early. [@Jesse] (2026-04-07 — fresh) → source
  • IT departments will gate-keep the hell out of AI adoption. However, if productivity gains are real, management will force it through. We’re seeing a classic Gartner Hype Cycle near the peak, bound to fall into the trough of disillusionment by year end. [@Will B] (2026-04-07 — fresh) → source
  • Questions how long insurers can hide and kick the can. They can’t let big insurers go bust or system collapses. Some similarities to 2008. Jamie Dimon saying similar things. [@Gaetan Warzee] (2026-04-07 — fresh) → source
  • The important view is stress appeared first in first-order holders (BDC vehicles, alt managers), and life insurers are the next cockroach in the chain. Lots of unmarked-to-market assets in insurer portfolios. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • CEOs will need to balance cost cutting and trusting AI. Reputation risk probably trumps cost cutting. Sees software companies getting squeezed on margins by clients but making it up by using AI to cut their own costs. [@Gaetan Warzee] (2026-04-07 — fresh) → source
  • Please note when listening and reading news feeds that the terms ‘private equity’ and ‘private credit’ are not the same thing and cannot be used interchangeably. [@Mark Tetreault] (2026-04-07 — fresh) → source
  • Chart from Gavin Baker showing AI impact on revenue/cost ratio is super important - next time someone says AI isn’t having material impact pull this out. [@Will B] (2026-04-07 — fresh) → source
  • Main concern is marking your own homework in private markets - PE commercial property valuations vs public markets show this risk. [@Stuart Hardy] (2026-04-07 — fresh) → source
  • Looked for private credit stress indicators like HYG/LQD but no good denominator exists to isolate credit spread from market beta. Real stress will show up in defaults and HYG eventually. [@thibault] (2026-04-07 — fresh) → source

Counter-arguments & data gaps

What would change this view

Events reckoned with

  • SightBringer ‘SaaS Fragility Map’ framework shared — reckoned 2026-04-02
  • iShares software ETF (IGV) down ~17% YTD and 26% from late-2025 highs — reckoned 2026-03-10
  • Goldman Sachs report on AI reshaping software sector without killing it — reckoned 2026-03-10
  • Jared buying puts on KIE insurance ETF on chatter about insurers loading up on private credit — reckoned 2026-02-25
  • Tech, comms, financials, discretionary down YTD while industrials, energy, staples, materials outperforming — reckoned 2026-02-25
  • Blue Owl Capital fails to secure ~$4B debt financing for CoreWeave data center project — reckoned 2026-02-20
  • Blue Owl Capital fails to secure ~$4B debt financing for CoreWeave data center - lenders cite AI infrastructure overcapacity risks — reckoned 2026-02-20
  • Software forward PE compressed from 35 to 19 — reckoned 2026-02-08
  • Anthropic releases 11 agentic AI plugins demonstrating AI can replace entire software workflows — reckoned 2026-01-30
  • Anthropic releases 11 agentic AI plugins demonstrating AI can replace entire software workflows — reckoned 2026-01-30
  • Anthropic releases 11 agentic AI plugins demonstrating AI can replace entire software workflows — reckoned 2026-01-30
  • Anthropic agentic AI plugins release triggers violent software repricing — reckoned 2026-01-30
  • Anthropic releases 11 agentic AI plugins demonstrating AI can replace software workflows — reckoned 2026-01-30