Rohit Krishnan makes a provocative claim that cuts through the usual anxiety surrounding artificial intelligence: advertising in AI isn't just a possibility, it is an economic necessity. While the prevailing wisdom treats ads as a corruption of the user experience, Krishnan argues they are the only viable mechanism to solve the impossible math of pricing intelligence for everyone from casual users to trillion-dollar corporations. This is a rare, clear-eyed look at the business model that will likely underpin the next decade of computing.
The Commoditization Trap
Krishnan begins by dismantling the idea that model superiority will remain a permanent competitive moat. He observes that while companies like OpenAI, Anthropic, and Google's Gemini have carved out niches, the underlying technology is rapidly becoming interchangeable. "Anything they produce also seems to get copied (and made open source) by Bytedance, Alibaba and Deepseek," he notes, highlighting how quickly proprietary advantages evaporate in this sector. The author correctly identifies that product features, such as OpenAI's Operator or Gemini's Storybook, are merely orchestration layers that competitors can replicate almost immediately.
This analysis is sharp because it shifts the focus from "who has the smartest brain" to "who can build the best software wrapper." Krishnan writes, "This isn't a fundamental difference in the model capabilities after all, it's a difference in how well you can create an orchestration." He argues that without a unique business model, these companies are destined to become low-margin utilities, akin to Amazon or Costco, rather than the high-margin software giants investors expect. Critics might argue that brand loyalty and network effects can sustain premium pricing longer than Krishnan predicts, but the rapid price drops of 95 to 99% in the last few years suggest his bearish view on model margins is well-founded.
"Unless the barrier simply is investment... the rise in usage will continue but if you're losing a bit of money on models you can't make it up in volume."
The Pricing Paradox
The core of Krishnan's argument rests on a fundamental economic problem: how do you charge a flat subscription fee when the value provided to different users varies by orders of magnitude? He illustrates this with a striking example of an executive using AI for high-stakes strategy versus a parent using it for bedtime stories. "If Elon Musk is using Claude to have a conversation, the answer to which might well be worth trillions of dollars of his new company," Krishnan writes, "that would be grossly underpaying you for the privilege of providing him with the conversation."
He posits that a flat subscription model fails to capture this value disparity, leaving massive revenue on the table or pricing out the casual user. The only historical solution to this differential pricing problem is advertising. Krishnan calculates the math, noting that with current token costs and ad click-through rates, the economics are tight but feasible, especially as inference costs continue to plummet. "Ads have an industry mean CPC (cost per click) of $0.63," he points out, suggesting that as the cost of generating text drops, the margin for ad-supported models widens significantly.
This is a compelling reframing. It moves the conversation from "ads are annoying" to "ads are the only way to make AI accessible to the masses while extracting value from high-end users." However, the argument glosses over the user experience friction. While Krishnan suggests AI could make ads less intrusive by being more relevant, the sheer volume of commercial intent in a conversational interface could fundamentally alter the trust users place in the system.
The Future of AI Commerce
Krishnan envisions a future where the interface itself becomes the marketplace. He describes a scenario where the AI doesn't just answer questions but facilitates transactions, with "Direct purchase links to products or links to services" appearing naturally within the flow of conversation. He predicts, "The conversion rates are likely to be much (much!) higher than even social media, since this is content, and it's happening in an extremely targeted fashion."
This vision is both exciting and unsettling. Krishnan acknowledges the fear that ads might corrupt the model's output, making it sycophantic or biased toward advertisers. He dismisses this as a technical hurdle rather than a fatal flaw, arguing that "we're still in the realm where we can't tell the model to not be sycophantic successfully for long enough periods of time." He believes the market will demand high-quality, helpful answers, forcing providers to keep the core intelligence clean even if the surrounding layer is commercial.
"The future, whether we want it or not, is going to be like the past, which means there's no escaping ads."
He concludes that the companies with the most user mindshare, specifically OpenAI and Gemini, are best positioned to execute this model because their interfaces are visual and conversational, unlike the command-line focus of some competitors. The implication is that the "free" tier of AI will become the dominant mode of interaction, funded by a sophisticated, AI-generated ad ecosystem that targets not just humans, but potentially other AI agents.
Bottom Line
Krishnan's strongest contribution is his ruthless application of basic economics to the AI hype cycle, demonstrating that the current subscription model is mathematically unsustainable for a utility that scales infinitely. His biggest vulnerability is underestimating the psychological toll of commercializing a tool users currently view as a neutral, helpful assistant. The reader should watch for the first major rollout of ad-supported tiers, which will likely serve as the stress test for whether users will accept this new reality or abandon the platform for open-source alternatives.