In his latest presentation on tech trends, ‘AI eats the world’ , Benedict Evans positions Generative AI as a “platform shift” – comparable to PC, Web, & Mobile disruptions. His market-level diagnosis is spot-on: the market-level scale is massive, $400bn+ annual capex, 800M weekly ChatGPT users, all innovation gravitating to AI, incumbent vulnerability.
But I’d respectfully push back on the terminology. The term “platform shift”(seems to be used as a catch-all) obscures what’s actually happening. What we’re seeing feels more like a transformative technology wave where existing platforms strengthen through AI integration. These are related but distinct dynamics.
A “platform shift” implies a new platform layer with network effects and switching costs, like iOS and Android created. But AI models don’t have these characteristics:
- Network effects? More ChatGPT users don’t make it better for other users. (In contrast, more Google searchers improve results for everyone.)
- User lock-in? Users freely substitute between ChatGPT, Claude, and Gemini weekly.
- Ecosystem control? Developers don’t find users on ChatGPT; they find ChatGPT on existing platforms (browsers, apps, websites).
What’s actually happening:
Existing platforms (Google Search + AI, Microsoft Office + Copilot, Facebook + AI recommendations) are integrating AI as a modularized supplier. That’s where the competitive moats are forming. I would rather call it market reorganization driven by transformative technology – not a new platform layer created by AI builders.
Why this distinction matters:
- For competitive strategy: Advantage comes from platform integration, not model quality. AI builders are in the supplier business; platforms own the moats.
- For organizational adoption: This explains why “having an AI strategy” is so hard. You’re not moving to a new platform; you’re integrating a transformative technology into existing ones.
- For investors: Platform businesses command premiums because of network effects and lock-in. AI infrastructure is transformative but has neither. Premium valuations belong to platforms that integrate AI well, not to model makers.
The real platform shifts are happening around AI, with existing platforms strengthening through AI not in AI itself.
What would you call it – platform shift or market reorganization or something else?