V. AI for investors
Is AI going to work out?
Not without structural changes.
What says AI must change?
Data. Rates of adoption went flat almost a year ago for LLMs across the board.
This is true at both at the consumer level (ChatGPT) and at the enterprise level (Microsoft, Salesforce).
Why did the rate of adoption go flat?
Users do not trust AI because it lies to them.
That can’t be true. How can you say it “lies?”
AI never says “I don’t know.” It is designed to provide an answer, even when it doesn't have one.
When an LLM lacks the information requested, it fabricates an answer – a “hallucination” – and delivers it in a confident, authoritative voice. Google Search never did this.
But AI models keep getting better and better!
New models have new features, but features have no impact on hallucinations.
By design, LLMs cannot remember everything – which requires them to fabricate “facts” to bridge gaps.
And by design, LLMs have no means of double-checking their “facts” before they respond. Google’s Gemini 3 hallucinates 88% of the time.
But many people say they use AI at least once a week or every day. Doesn’t that count as adoption?
No. They are replacing Google with AI for search.
That is merely migration of an audience from one platform to another. The net effect on total audience is zero.
So what? They’re still using it!
- Not for collaboration.
- Not for higher-level reasoning.
- Not for “genius-level” pattern-matching recognition.
- Not to cure cancer.
They are using it to find recipes, restaurant reviews and settle bar bets. That’s retrieval, not cognition. Retrieval has been free for decades, and users will not pay for it – because they never paid for Google.
But people are paying for AI, right?
Very few. Roughly 95% of ChatGPT users do not pay – and why should they? They are merely using it to replace Google.
But AI can be monetized through advertising if enough people use it, right?
Nope. Scale does not solve the problem.
Here’s why:
Google is an advertising company. Not a search company.
Meta is an advertising company. Not a social media company.
- YouTube
- TikTok
- Yelp are all advertising companies.
They make money because ads are delivered at extremely low marginal cost.
So why can’t AI do the same thing?
Because AI is not search or social media. It is cognition, which requires much more powerful chips, more energy and more money. And the cost per query does not decrease as the number of queries increases, because the cost of each query is fixed – for now.
When advertising migrates to AI, costs explode. If search ads move from Google to its LLM, Gemini, Google keeps the revenue but must now pay LLM inference costs that are orders of magnitude higher than search. So profit margins shrink.
Same problem for Meta if ads move from Facebook, Instagram and WhatsApp to its LLM called LLaMA.
Could AI just embed ads directly in responses?
No.
Ads in search and social media are adjacent and are not part of the content. But AI responses are delivered in a confident, authoritative voice – just like a real person – Google never “talked to you” directly like AI does.
And unlike Google, AI will seem more personal when users switch from text to voice. But users will not trust their medical, financial, legal or personal information with an advertising-driven AI system that sounds like a trusted friend while recommending its advertisers’ products and services.
So is AI doomed as a business?
Yes.
No trust → no adoption → no scale → no profit from subscriptions or advertising.
Is there a solution, or is AI the next Segway?
Yes.
AI 2.0 fixes the economics and the trust problem.
What does AI 2.0 do differently?
- Reduce capital expenditures and operating expenses – by up to ~90% – via pre-execution provisioning.
- Charge based on actual compute used, not tokens or flat-rate plans.
- Provide 100% loss-less memory, eliminating hallucinations.
- Provide user-level governance for control and customization.
So, when does this happen?
When investors, LLMs and the general public face the facts and stop believing in a narrative not grounded in reality.