⁠Why OpenAI and Anthropic Won't Win the App Layer | Glean Founder

Even when models can run on‑prem, geopolitical trust blocks adoption because companies fear hidden backdoors or political risk from models originating in adversarial jurisdictions.

Enterprise AI usage follows a power law because a few teams experiment intensively and consume most tokens while the majority stick to basic summarization, concentrating costs and advanced use cases.

Budget pressure is driving open model adoption mainly because contractual assurances have reduced training‑on‑customer‑data fears, leaving cost and rising open model quality as the immediate drivers.

Generated code floods repos quickly, yet automated code is often opaque and brittle, so removing human review will degrade long-term correctness, security, and maintainability.

Short-term inference prices rose to prove commercial viability, but historical tech trends mean token costs are likely to fall over time, which will squeeze model providers' margins and force new business models.

Per-person output will rise with AI, but firms will raise expectations accordingly, so companies may still hire to scale production and capture market share rather than shrink headcount.

AI lets companies merge specialties into composite roles because automation handles many specialized subtasks, allowing one person to cover engineering, product, design, or sales across a workflow.

Delivering AI value depends more on engineering around models because curated context, connectors, and preassembled data prevent models from burning tokens and time brute‑forcing the right information.