Introduction

Chatbots will not make big money from simple shopping links. The real jackpot is personalized ads built into AI chats.

What's the problem it solves?

The article asks a simple question: how should big AI chatbots like ChatGPT make real money long term.
Right now, things like affiliate links and tiny shopping fees will not cover the huge costs of running these models.
The author argues that this whole “agentic commerce with affiliate links” idea is weak, and that chatbots need a better, more scalable money engine.

That better engine, in his view, is personalized advertising.

Quick Summary

OpenAI has a giant user base and giant costs. ChatGPT has around 800 million weekly users, but only a small slice pay for the product. OpenAI is losing billions today while also planning to spend an eye watering amount on chips and infrastructure. To support a huge valuation and survive long term, it must grow revenue fast.

One new bet is Instant Checkout: ChatGPT suggests products from partners like Walmart and Etsy and takes a fee when people buy. This works like affiliate links inside the chat. The author says this is the wrong main model for chatbots. It only earns money on a narrow set of “I want to buy something” queries, depends on product feeds and reviews from other sites, and conflicts with powerful players like Amazon.

He argues that personalized ads are a much better fit. Chatbots can use more than just the current message to decide what to show. They can draw on rich behavior data and serve ads across many formats, including video, not just text. Ad auctions let advertisers bid based on what each user view is worth to them, which tends to maximize both platform revenue and the total value created. Affiliate systems cannot do this. They mostly chase high conversion rates, often pushing cheaper products, which lowers total value.

On top of that, if chatbots take over the last step of the purchase, they can quietly kill the review and content ecosystem they depend on. Bloggers and reviewers earn money from their own affiliate links. If chatbots take the traffic and the transaction, those sites may die, and the bot loses a key input: real human reviews. Retailers also have mixed incentives. Amazon, whose profits depend heavily on its own ads and direct customer data, is already blocking third party agents and suing those that try to buy on its platform. That alone may cap how far affiliate style chatbot commerce can go.

So the author thinks the future of chatbot monetization looks less like “shopping agents” and more like “super advanced ad platforms” that route existing demand to the best products using personalized ads, not simple affiliate referrals.

Key Takeaways

• OpenAI needs far more revenue than subscriptions and small affiliate fees can give, given its huge losses and planned spend.
• Affiliate style “agentic commerce” inside chatbots is narrow, fragile, and depends on data and reviews from others.
• Three bad assumptions drive the current hype: that only chat context can be used to target, that chat will stay text only, and that review data will always be there.
• Chatbots can and will be multi modal, often closer to YouTube than Google Search, which fits an ad model much better.
• Personalized ads use auctions where advertisers bid for attention, which tends to maximize value of each impression for everyone.
• Affiliate models only make money on commercial queries and often favor low price, high conversion items, which can shrink total revenue.
• Heavy use of affiliate style agents clashes with key players like Amazon, who want to keep control of the customer and the ad money.
• Instant Checkout does not even give merchants a true direct customer relationship, since they cannot use those emails for marketing.
• The author expects personalized advertising to become the main money engine for ChatGPT and chatbots in general.

What to do

• If you build chatbots or AI products, plan for a serious ad model, not just affiliate links and checkout buttons.
• Think of your chatbot as an ad surface that routes existing demand to good products, not as a magical sales agent that creates demand.
• Start building your own rich first party data so you can support personalized ads without depending only on in chat keywords.
• Design answers so that ads can fit in without breaking trust: clear labels, real recommendations, and useful context.
• Do not rely fully on external review sites that you might also be draining of revenue. Find ways to support or partner with them.
• If you are a retailer or marketplace, decide where you stand: will you let third party chatbots sit between you and the customer or run your own agent.
• If you run a brand, expect AI chatbots to become key ad channels. Prepare creative, measurement, and bids for this new surface early.
• For investors and operators, judge AI businesses not just by user growth, but by whether their monetization model can realistically match their infra costs.

‍