a16zAI Is Coming For These 3 Industries In 2026 (a16z Big Ideas)
CHAPTERS
Big Ideas 2026: Three AI-driven shifts across industry, finance, and enterprise software
Erik Torenberg frames the episode as part two of a16z’s 2026 Big Ideas, featuring three investors’ forecasts. The throughline is AI and software reshaping foundational layers of the economy: how we build things, how money moves, and how work gets done inside companies.
- •Electro-industrial stack as the foundation of industrial competitiveness
- •Financial services/insurance reaching a legacy-replacement tipping point
- •Dynamic agent layer emerging to challenge systems of record
- •Perspectives come from investors actively backing these transitions
The electro-industrial stack: electrified components become the new industrial core
Ryan McEntush introduces the “electro-industrial stack” as the next industrial evolution happening inside the machines themselves. He ties together EVs, drones, data centers, and modern manufacturing around shared building blocks like batteries, power electronics, compute, and motors.
- •Electro-industrial stack links multiple categories via common components
- •Industrial progress shifts from factory-level changes to machine-level internals
- •Embodied, electrified hardware becomes the channel through which software/AI impacts the physical world
- •Winning requires deep technical expertise in hard-to-build components
America vs. China: technology parity, ecosystem gap
McEntush argues the U.S. can match China’s industrial technologies, but the larger challenge is scaling them economically. China’s advantage is the density and speed of its end-to-end industrial ecosystem—suppliers, materials, and enabling institutions.
- •U.S. can do the tech (e.g., rare-earth processing), but scaling at low cost is harder
- •Key gap: mature tier-1/2/3 supplier networks and rapid coordination mechanisms
- •U.S. leaders often vertically integrate “by necessity, not strategy”
- •Bottlenecks move unless the broader ecosystem grows alongside core innovation
How to build it in the U.S.: blend talent, co-locate, and rebuild industrial prestige
To compete, McEntush emphasizes combining Silicon Valley software velocity with industrial veterans’ domain knowledge. He also stresses tight coupling between engineering and manufacturing and elevating the mission to attract top talent.
- •Blend software talent/culture with experienced industrial operators
- •Leverage “what’s been tried before” from legacy aerospace/industrial expertise
- •Co-locate engineering and manufacturing to accelerate design-for-manufacturing loops
- •Create prestige/purpose to pull elite talent toward industrial problems
Supply chains as strategic leverage in an AI-powered world
McEntush closes by emphasizing that reshoring and owning supply chains for key components will determine economic and military power over the long run. As AI increases automation and industrial capability, control over these inputs becomes more consequential.
- •Critical components: batteries, power electronics, compute, motors
- •Reshoring/vertical integration becomes necessary for reliability and speed
- •AI-enabled automation increases the strategic value of supply chain control
- •Long-term national competitiveness ties to industrial talent base and supply chain ownership
Financial services & insurance tipping point: legacy risk now exceeds change risk
Angela Strange predicts a dramatic turning point where major institutions let legacy contracts lapse and adopt AI-native competitors. The catalyst is new infrastructure that unifies data across cores and external/unstructured sources into a new system of record.
- •Institutions begin replacing entrenched legacy vendors rather than extending contracts
- •AI-native platforms unify fragmented data into a new system of record
- •Replatforming becomes less risky than staying on brittle, outdated systems
- •The shift is positioned as already starting among major incumbents
What changes for operators: parallel workflows, expanded platforms, and bigger winners
Strange outlines three major impacts once AI-native infrastructure takes hold. Workflows become parallelized, risk/compliance categories converge into broader platforms, and the biggest winners become much larger by absorbing labor and expanding category scope.
- •Parallelized workflows reduce swivel-chair work (less copy/paste across screens)
- •Example: mortgage underwriting tasks can run in parallel with agent assistance
- •Risk platforms expand by unifying onboarding/KYC/KYB/monitoring/support signals
- •Winners can be “10x bigger” as software captures labor and expands TAM
Why now: mainframes near limits, AI revenue upside, and credible AI-first vendors
Strange argues the timing is different now because legacy cores are strained, AI creates immediate upside, and the vendor landscape finally includes viable, re-architected AI-first platforms. Entrepreneurs who are both deeply technical and domain-native are building real replacements.
- •Many FIs still run on decades-old mainframes approaching scaling limits
- •AI unlocks revenue (e.g., insurance underwriting throughput) once documents/data can be processed faster
- •New platforms are built to scale and remain flexible as AI evolves
- •Competitive pressure: “your competitors using AI” becomes the real threat
Early adopters gain compounding advantages through unified data
Strange highlights how early adopters can develop reputations as forward-thinking partners and quickly outpace slower rivals. Unified data layers enable better customer experiences and operational leverage, translating into large margin improvements in some business lines.
- •Early adoption can reorder winners/losers among incumbents
- •Some firms move from low-margin to high-margin operations (e.g., mortgage servicing)
- •Unified data prevents redundant marketing and fragmented customer support
- •Agents + unified context enable proactive, cross-product customer experiences
Call to founders: modernize banking/insurance plumbing in 2026
Strange ends with a builder-focused message: the market is ready and the opportunity is enormous for AI-first infrastructure. Founders with deep curiosity about “archaic” processes can build faster than ever and sell into urgent demand.
- •Large, ready-to-buy market as institutions accept replatforming
- •Opportunity targets legacy/archaic workflows and systems
- •AI-first platforms can be built and iterated faster than prior generations
- •Demand accelerates as competitive gap widens between adopters and laggards
Systems of record under threat: dynamic agents collapse intent-to-execution
Sarah Wang predicts systems of record will lose primacy as agents can execute tasks directly from user intent. She frames this as the first credible 10x disruption after prior SaaS waves failed to dislodge systems of record with UI improvements alone.
- •Systems of record historically win due to data gravity and stickiness
- •Prior SaaS challengers largely failed by competing on UI alone
- •Agents collapse the distance between intent and execution, enabling 10x experiences
- •A new “dynamic agent layer” becomes the primary interface for employees
ITSM as the concrete wedge: from tickets to near-instant fulfillment
Using IT service management, Wang illustrates how agents can transform slow, form-based workflows into rapid execution. Advances in LLMs let systems interpret requests, route them to workflows, and complete actions reliably within existing stacks.
- •Legacy ITSM (e.g., ServiceNow category) is poised for workflow disruption
- •Agents can extract intent, classify requests, map workflows, and identify entities
- •Example: requesting software access shifts from slow ticketing to near-instant resolution
- •Trust and reliability are prerequisites—agents must be accurate to be adopted
Where value accrues: foundation models matter, but the agent layer compounds
Wang distinguishes the enduring value of the foundation model layer from the compounding advantage of the agent layer closest to the user. The agent layer captures user context and preferences, which can become defensible over time.
- •Foundation models remain valuable infrastructure
- •Agent layer sits closest to users and continuously learns preferences/context
- •Data gathered through interactions becomes an advantage for agent-native products
- •The interface layer shifts from records to action-oriented execution
Competitive dynamics: fast iteration rewards new entrants, 2026 as the crossover year
Wang argues the market is moving at weekly/daily improvement cycles, favoring teams that ship quickly and deliver reliable outcomes. She points to early examples of agent-native tools outperforming agents bolted onto iconic platforms, and predicts 2026 is when agents overtake systems of record.
- •Rapid product improvement cycles create openings for startups
- •Reliability bridges intent-to-execution; without it, customers won’t trust agents
- •New AI SRE/ops players can beat agent add-ons on legacy platforms
- •Prediction: 2026 is the year the dynamic agent layer overtakes systems of record
