The Siege Curve: How Disruptive Technologies Are Really Adopted (Zenodo (CERN European Organization for Nuclear Research))

Summary

• Gartner forecast generative AI spending of $644 billion for 2025, even as OpenAI reported 800 million weekly ChatGPT users in October 2025.p.1
• Microsoft, Google and OpenAI all offered copyright indemnities to customers in 2023 long before any landmark judicial ruling.p.5
• Amazon bought Quidsi for $545 million in 2010 after price matching strategies, while OpenAI asked investors in its $6.6 billion round not to back competitors in 2024.p.8
• Uber endured an $8.5 billion loss in 2019 during its regulatory siege before reaching its first annual operating profit of $1.1 billion in 2023.p.15
• Global start-up funding reached $297 billion in the first quarter of 2026, though roughly four-fifths of that capital went to fewer than ten companies.p.17
• Major technology companies planned between $650 billion to $700 billion of AI capital expenditure for 2026.p.17

Key insights

1. The difficult middle period of disruptive technology adoption is a structural siege rather than a subjective trough of disillusionment.p.1
2. The Siege Curve tracks four distinct lines comprising consumer usage, business adoption, capital, and the field of viable firms.p.3
3. A siege forms a two-sided squeeze where external legal, political, and public backlash collide with internal shortages of skills, scale, and infrastructure.p.5
4. Capital moves ahead of usage during a siege, concentrates in front-runners, and reprices once landmark regulatory or legal risks are priced.p.7
5. Firm entry peaks near the first challenge gate when building costs are dropping and permission costs remain minimal.p.9
6. Empirical studies indicate market pioneers fail 47% of the time, compared with an 8% failure rate for early market leaders.p.11
7. Surviving firms assemble drawbridges using patents, rules, talent, capacity, and deep capital pools created from the same fronts that besieged them.p.12
8. The decisive US landmark fair use ruling or federal statute for generative AI is predicted to occur between 2027 and 2030.p.18