Accel Report: AI Infrastructure Race Demands $7 Trillion by 2030

The Race for Compute Defines Next Five Years

Accel's newly released 2025 Globalscape report positions the period through 2030 as a "race for compute" requiring approximately $4 trillion in capital expenditure to build 117 gigawatts of new AI data center capacity. This electricity demand equals the combined consumption of Italy, Spain, and the UK. The buildout depends on generating $3.1 trillion in data center revenue by 2030, equivalent to increasing global GDP growth by 1-2 percentage points annually.

Market Power Consolidates Around "Super Six"

Six technology companies (Apple, Microsoft, Google, Amazon, Meta, and Nvidia) now command 50% of NASDAQ market capitalization, collectively worth $20.7 trillion. These firms added $4.9 trillion in market value over the past year alone while maintaining 30% average revenue growth. This concentration reflects how AI infrastructure investments have become asymmetric advantages for companies with existing scale and capital, creating unprecedented market dominance.

AI-Native Applications Reach Scale at Record Speed

A new generation of AI-native companies is achieving commercial milestones faster than any previous software era. Lovable reached $100 million ARR in just eight months, while Cursor hit $50 million ARR within 30 months of founding. These companies benefit from self-serve distribution models and developer-led adoption that bypass traditional enterprise sales cycles, compressing the journey from $1 million to $100 million ARR into 2-3 years versus the historical 7-10 years.

Geographic Divergence Shapes Investment Landscape

While the US dominates foundation model funding with massive rounds for OpenAI, Anthropic, and xAI, European and Israeli companies demonstrate competitive strength in AI applications. EU/IL funding for cloud and AI applications (excluding models) represents roughly two-thirds of US levels. Top European rounds include Helsing at $683 million and Cyera at $540 million, comparable to US application-layer financing, suggesting geographic specialization rather than across-the-board US dominance.

Energy Infrastructure Emerges as Critical Constraint

Despite abundant capital availability, electricity infrastructure may become the binding constraint on AI expansion. The US faces a projected 36-gigawatt electricity shortfall relative to data center buildout plans. Capital intensity is concentrated among five hyperscalers (Amazon, Microsoft, Google, Meta, Apple) with combined 2025 capex of $383 billion, representing 85% of total forecast spending, but power generation and grid distribution capacity must keep pace for infrastructure plans to materialize.

Five Trends Define Next AI Wave

Accel identifies five emerging categories shaping 2025-2030 development: AI security addressing new attack surfaces in agentic systems; vertical-specific AI applications automating industry workflows in healthcare, legal, and construction; enterprise deployment of autonomous agents and computer-use models; AI-driven code generation ("vibe coding") reshaping software development; and voice and synthetic media becoming standard enterprise UX. Early enterprise agentic use cases show compelling ROI, with companies like UiPath achieving 98% automation rates and Decagon delivering 80% ticket deflection.