a16zAI Is Coming For These 3 Industries In 2026 (a16z Big Ideas)
CHAPTERS
- 0:00 – 0:35
Big Ideas 2026 overview: three AI-driven inflection points
Erik Torenberg tees up three investor theses for 2026 across American dynamism, financial services/insurance, and enterprise software. The episode frames these not as distant predictions but as shifts already underway, with implications for founders and incumbents alike.
- •Three themes: electro-industrial stack, AI-native financial infrastructure, dynamic agent layer in enterprise software
- •Focus on 2026 as an acceleration year rather than a far-future horizon
- •Investor perspective: where platform shifts create new category winners
- 0:35 – 1:06
The electro-industrial stack: electrified components become the new industrial foundation
Ryan McEntush argues the next industrial evolution happens “inside the machines” via a shared stack of electrified components. Batteries, power electronics, compute, and motors become the common backbone across EVs, drones, data centers, and manufacturing.
- •Electro-industrial stack as a cross-industry enabling layer
- •Embodied/electrified components as the route by which software impacts the physical world
- •Scope includes EVs, drones, data centers, modern manufacturing
- 1:06 – 1:36
America vs. China: technology parity vs. ecosystem disadvantage
McEntush pushes back on simplistic narratives about China being uncatchably ahead. He contends the U.S. can match much of the core technology, but China’s advantage is the surrounding industrial ecosystem that makes scaling fast and cheap possible.
- •U.S. capability: strong engineering and ability to execute known processes (e.g., rare earth processing)
- •True constraint is industrial-scale ecosystem buildout and cost competitiveness
- •Ecosystem depth (tiered suppliers + institutions) is China’s structural edge
- 1:36 – 2:06
Vertical integration by necessity: when supplier networks can’t keep up
Using SpaceX and Anduril as examples, McEntush explains that vertical integration often happens because the supplier base isn’t mature enough to scale alongside fast-moving primes. In China, dense supplier tiers reduce the need for such self-reliance.
- •Vertical integration as a response to missing scalable suppliers
- •China’s tier 1/2/3 supplier networks enable speed
- •U.S. catch-up requires years/decades of ecosystem development
- 2:06 – 2:37
Building the U.S. industrial talent engine: blend software speed with veteran know-how
To build the electro-industrial stack domestically, teams must combine Silicon Valley software culture with deep industrial and aerospace experience. McEntush emphasizes learning from what’s been tried before while moving with modern software velocity.
- •Blend software talent with industrial veterans (e.g., propulsion/aerospace heritage)
- •Institutional knowledge reduces repetition and accelerates execution
- •Modern software practices can unlock faster iteration in industrial contexts
- 2:37 – 3:37
Co-locating engineering and manufacturing + making the mission prestigious
McEntush argues that tighter integration between design and production speeds iteration, especially via design-for-manufacturing practices. He also highlights the need to build prestige and purpose to attract top-tier talent that has many competing options.
- •Co-location enables faster feedback loops and design-for-manufacturing
- •Speed comes from integration on the same footprint or within the same ecosystem
- •Talent competition requires mission, prestige, and purpose—not just compensation
- 3:37 – 4:22
Supply chains as strategy: the 21st-century economic and military lever
The thesis culminates in the claim that owning and reshoring key electro-industrial supply chains will determine national and corporate winners. As AI increases automation and industrial capability, supply chain control becomes more decisive for power.
- •Critical components: batteries, power electronics, compute, motors
- •Reshoring/vertical integration becomes essential as AI-driven automation rises
- •Supply chain ownership influences future economic and military strength
- 4:22 – 4:52
Financial services turning point: legacy replacement becomes less risky than standing still
Angela Strange predicts 2026 brings a tipping point where the danger of maintaining legacy cores exceeds the risk of migrating. Institutions will increasingly let old contracts lapse and adopt AI-native competitors built for unified data and scale.
- •Risk calculus flips: not changing becomes the bigger risk
- •Incumbents begin replacing long-standing vendors
- •AI-native cores unify legacy + external + unstructured data into a new system of record
- 4:52 – 6:24
What unified AI-native infrastructure changes: parallel workflows, bigger platforms, 10x winners
Strange outlines three structural impacts of modern platforms: parallelized work, expanded categories via unified risk/customer data, and larger software winners. AI absorbs manual labor and unlocks scale, making category boundaries blur and TAMs expand.
- •Parallelization replaces screen-hopping and manual copy/paste
- •Unified platforms combine onboarding/KYC/KYB/monitoring/service interactions into risk views
- •Winners grow 10x as software consumes labor and expands category scope
- 6:24 – 7:25
Why now: mainframes at the breaking point, AI revenue upside, and credible new entrants
She argues the timing is different because old systems are nearing failure under scale, AI makes the opportunity cost visible, and new vendors are finally viable. Founders are re-architecting platforms to be flexible for AI today and future iterations.
- •Legacy cores (often mainframes) are strained and brittle at modern scale
- •AI reveals lost revenue (e.g., insurance underwriting throughput constraints)
- •Entrepreneurs are rebuilding platforms for scalability and AI extensibility
- 7:25 – 7:55
Early adopters pull ahead: reputation effects and margin expansion from new platforms
Strange describes how early adopters gain both operational advantages and market perception as forward-thinking partners. She points to dramatic margin expansion in certain functions and warns that laggards may need years to catch up.
- •Platform adoption creates reputational advantage with partners and vendors
- •Examples of margin expansion (e.g., mortgage servicing) through operational leverage
- •Speed matters: multi-year catch-up windows can reshape competitive standings
- 7:55 – 9:23
Unified data enables “beautiful” customer experiences—and a call to founders
She connects infrastructure modernization to better end-user experiences, like eliminating redundant product marketing and enabling truly informed customer service. The segment closes with a founder call: archaic banking/insurance workflows are ripe for AI-first rebuilds, and buyers are increasingly ready.
- •Bad CX often stems from fragmented customer data silos
- •Unified data layer + agents can anticipate needs and personalize service end-to-end
- •Founder opportunity: rebuild legacy pain points faster, with higher customer readiness
- 9:23 – 10:24
Enterprise shift: systems of record lose primacy as a dynamic agent layer emerges
Sarah Wang predicts agents will erode the dominance of traditional systems of record by collapsing the distance between intent and execution. This creates a genuine 10x leap, unlike prior SaaS attempts that mainly improved UI over legacy incumbents.
- •Passive systems of record become less compelling when agents can execute intent
- •Prior ERP/SaaS challengers failed because UI improvements weren’t enough
- •Agents create a step-change (10x) by compressing intent-to-action cycles
- 10:24 – 10:54
ITSM as the wedge: why agent-based workflows can upend incumbents like ServiceNow
Wang uses IT service management to illustrate how agents can transform slow, ticket-based processes into near-instant fulfillment. With improved LLM capabilities, agents can parse requests, map workflows, identify entities, and execute reliably inside the enterprise stack.
- •ITSM poised for major change over the next five years
- •Agents extract intent, classify requests, map to workflows, and execute quickly
- •Value hinges on accuracy and reliability to earn user trust
- 10:54 – 12:29
Where value accrues: foundation models stay important, but the agent layer wins—and the race accelerates
Wang argues that while foundation models remain valuable, the agent layer closest to the user will compound advantage by capturing preference and interaction data. Rapid iteration speed becomes decisive, enabling new AI-native entrants to beat agents built atop established platforms.
- •Agent layer near the user accrues durable value via preference/interaction data
- •Products improve weekly/daily; fast-moving teams gain an edge
- •Examples of new AI SRE entrants outcompeting agent add-ons to legacy platforms
- •Prediction: 2026 is the year the dynamic agent layer overtakes systems of record
