Where Enterprises are Actually Adopting AI

· startups business · Source ↗

TLDR

  • 29% of the Fortune 500 and ~19% of the Global 2000 are live, paying customers of a leading AI startup as of early 2026.

Key Takeaways

  • Coding dominates enterprise AI adoption by nearly an order of magnitude; Cursor, Claude Code, and Codex are cited as hyper-growth examples with 10-20x engineer productivity gains reported.
  • Customer support is the second major category: clear SOPs, quantifiable metrics (tickets, CSAT, resolution rate), and existing BPO outsourcing lower adoption friction significantly.
  • Enterprise search is the third horizontal; Glean, Harvey (legal, ~$200M ARR within 3 years), and OpenEvidence (medical) each built large businesses around domain-specific information retrieval.
  • Legal and healthcare broke their historically slow software adoption patterns: Harvey hit ~$200M ARR, Eve reached a $1B valuation with 450+ customers, and Abridge/Ambience/Tennr grew rapidly on discrete clinical use cases.
  • The MIT claim that 95% of generative AI pilots fail to convert is directly contradicted by a16z’s aggregated private data from enterprise AI startups.

Why It Matters

  • Fortune 500 enterprises have historically taken years to sign a first startup contract; AI compressed that cycle to under three years from ChatGPT’s November 2022 launch.
  • Coding is upstream of all software, so AI’s exponential improvement there reduces the floor to build in every other domain while simultaneously compressing startup moats.
  • Legal and healthcare, two markets that resisted traditional workflow software, are adopting AI because the technology matches the actual work: parsing dense text, reasoning over large corpora, and handling unstructured tasks.

Kimberly Tan, Andreessen Horowitz · 2026-04-08 · Read the original

Related coverage