20VC with Harry Stebbings• Apr 20, 2026• 54:54Interview

Aaron Levie: Everyone is Wrong; We'll Have More Developers in 5 Years

Aaron Levie•Founder and CEO, Box

Executive Summary

AI's primary impact on the enterprise will be augmenting human workers and redesigning workflows, not mass job displacement.
New roles like 'agent operator' will emerge.
The US-China AI competition is primarily a commercial and economic race, not an existential one where a short-term lead is decisive.
The global adoption of US technology stacks is a key strategic advantage.
Enterprises will need to shift AI spending from constrained IT budgets to larger operational budgets (OPEX), treating AI as a tool to improve core business productivity, not just a technology expense.
The slow, complex reality of enterprise AI adoption, including regulatory hurdles and the need for new infrastructure like 'agent observability', presents a multi-decade opportunity that Silicon Valley often underestimates.

12 quotes

Concerns Raised

The tech industry is myopic and underestimates the complexity of enterprise AI adoption.
Enterprises face significant challenges in redesigning workflows for human-agent collaboration.
Regulatory, compliance, and security hurdles will slow the diffusion of AI technology.
The transition from traditional apprenticeship models may be difficult as AI automates junior-level tasks.

Opportunities Identified

Creation of 500k-1M new 'agent operator' jobs to manage and oversee AI systems.
A decade's worth of work in organizing enterprise data and redesigning workflows for AI.
A new category of essential infrastructure for agent observability and evaluation is emerging.
AI vendors can tap into massive operational budgets (OPEX) instead of being limited to smaller IT budgets.

Key Themes

AI's Impact on the Workforce

Contrary to fears of mass job displacement, AI is expected to augment human capabilities and create new roles. Speaker Aaron Levie argues that professions like engineering and law will see increased demand as AI tools lower the barrier to producing work, which still requires human review, oversight, and accountability.

This provides a counter-narrative to the common dystopian view of AI and jobs, suggesting that the primary challenge is not replacement but rather retraining, apprenticeship, and redesigning roles for human-agent collaboration.

The Realities of Enterprise AI Adoption

The diffusion of AI into large enterprises will be a slow, multi-year process, much like the cloud transition. Companies face significant hurdles in redesigning workflows, ensuring regulatory compliance, and managing security, which contrasts with Silicon Valley's expectation of rapid, frictionless adoption.

This highlights a massive, long-term market opportunity for companies providing the tools, services, and infrastructure needed to manage this complex transition, from agent observability to professional services.

The Geopolitics of AI

The AI race between the US and China is framed as a commercial and economic competition rather than an existential military one. The argument is that a one or two-month lead in model development is not a decisive, permanent advantage, and that US influence is best secured by having its technology stack power the global economy.

This nuanced perspective challenges the more alarmist national security narratives and suggests a strategy focused on commercial dominance and open innovation rather than purely defensive or restrictive measures.

New Economic Models for AI

AI introduces a new consumption-based cost model (tokens) that doesn't fit neatly into traditional enterprise IT budgets. The key shift will be funding AI initiatives from operational budgets (OPEX), tying the cost directly to productivity gains in specific business units like marketing or finance, which could double the total addressable market for technology vendors.

This signals a fundamental change in how technology is procured and valued within corporations, moving from a fixed-cost license model to a variable-cost, value-aligned model that requires new budgeting and ROI calculation methods.

Silicon Valley Myopia

A recurring point is that the tech industry is often insular and fails to appreciate how the other 85% of the economy operates. AI coding tools, for example, won't just make tech companies more efficient; they will empower non-tech companies in sectors like agriculture and pharmaceuticals to build their own powerful engineering capabilities for the first time.

This identifies a major blind spot in market analysis, suggesting that the largest long-term impacts and opportunities for AI may lie outside the traditional tech sector, fundamentally changing the distribution of technical talent and innovation.

Get started free

Topics

AI in the Enterprise Aaron Levie Box Job Creation Job Displacement Agent Operator AI Agents US-China AI Competition Geopolitics Jensen Huang NVIDIA Enterprise Software Workflow Automation Tokenomics IT Budgets OPEX Agent Observability Braintrust

Processed Apr 21, 2026 yt-dlp + mlx-whisper + Gemini