𝐓𝐡𝐞 𝐏𝐫𝐢𝐜𝐞 𝐎𝐟 “𝐀𝐈 𝐅𝐨𝐫 𝐀𝐥𝐥”

Despite widespread AI anxiety, AI is often touted as a superpower at everyone's fingertips. The early data supports that. AI-exposed industries are showing real productivity gains. Output is rising. Employment is holding. The dream, apparently, is coming true.

But look closer.

A recent Morgan Stanley report showed that a small tier of top performers, armed with AI, is already absorbing the work of entire layers below them. 90% may not be fired by AI. They may be displaced by the 10% who know how to wield it. The productivity boom may 𝐧𝐨𝐭 𝐥𝐢𝐟𝐭 𝐚𝐥𝐥 𝐛𝐨𝐚𝐭𝐬 equally. It may concentrate leverage.

Add the sustainability issue of current 𝐀𝐈 𝐩𝐫𝐢𝐜𝐢𝐧𝐠.

Twenty dollars a month buys access to tools that would have looked like science fiction three years ago. But that price does not reflect the full cost of frontier AI. For now, the gap is covered because your prompts, edits, and corrections are training data. Within three to five years, synthetic data may replace that human input. The subsidy ends. Pricing resets.

The democratization was real. But 𝐭𝐡𝐞 𝐦𝐨𝐝𝐞𝐥 𝐰𝐚𝐬 𝐬𝐮𝐛𝐬𝐢𝐝𝐢𝐳𝐞𝐝.

If that subsidy fades, the consequences are predictable. Large enterprises can absorb six- and seven-figure AI contracts. Well-funded startups and wealthy professionals can pay up. Small businesses, public-sector teams, independent workers, schools, and developing economies may not. What looked like a level playing field may have been an introductory offer.

This is a scenario, not a certainty. 𝐓𝐡𝐫𝐞𝐞 𝐞𝐱𝐢𝐭𝐬 exist. None are perfect.

𝐓𝐡𝐞 𝐟𝐢𝐫𝐬𝐭 𝐢𝐬 𝐫𝐞𝐠𝐮𝐥𝐚𝐭𝐢𝐨𝐧. If AI becomes essential cognitive infrastructure, countries may treat access the way they treated electricity, broadband, and essential medicines. How to regulate without slowing innovation is a delicate act.

𝐓𝐡𝐞 𝐬𝐞𝐜𝐨𝐧𝐝 𝐢𝐬 𝐭𝐡𝐞 𝐂𝐡𝐢𝐧𝐞𝐬𝐞 𝐨𝐩𝐭𝐢𝐨𝐧. DeepSeek showed frontier-adjacent performance at a fraction of Western costs. Shopify replaced its OpenAI pipeline with Alibaba's Qwen 3, cutting per-unit AI costs by 75x. If Western AI prices itself beyond reach, the world will not go without. It will route around, which raises its own questions.

𝐓𝐡𝐞 𝐭𝐡𝐢𝐫𝐝 𝐢𝐬 𝐚𝐝𝐯𝐞𝐫𝐭𝐢𝐬𝐢𝐧𝐠. In January 2026, Sam Altman, who once called ads in AI "uniquely unsettling" and a "last resort", began testing sponsored content in ChatGPT's free tier. Many people want AI and don't want to pay. An ad-supported model could keep the door open. Search ads shaped what we clicked. Social ads shaped what we believed. AI ads may shape what we ask and which answers feel natural.

The question is not whether "𝐀𝐈 𝐟𝐨𝐫 𝐚𝐥𝐥" survives. It probably will.

The question is what it costs to survive, and whether we can live with that.

#DigitalDivide #TechPolicy #AIAccess #AIGeopolitics #AdSupportedAI

Contact

bruno.gentil@sherpaconsultingasia.com

© 2026. All rights reserved.