For close to three years, generative AI has been sold as the biggest technological revolution since the internet, expected to have profound effects on humanity. US tech leaders have poured hundreds of billions of dollars into AI infrastructure, governments are racing to build national AI capabilities and hubs, and investors continue to reward companies promising an AI-powered future, evident from the massive stock rallies.
However, one of the industry’s most vocal critics believes the narrative is starting to crack.
During a recent CNBC appearance, technology analyst Ed Zitron argued that generative AI’s underlying business model simply doesn’t work. He even claimed that OpenAI is potentially pushing their IPO to 2027 because it couldn’t get a $1 trillion valuation. The company reportedly recorded $38.5 billion in losses last year on revenue of $13.07 billion.
His comments quickly spread across online AI communities, where they reignited a debate that has been quietly growing: is AI experiencing a genuine technological revolution, or is it riding an unsustainable investment bubble?
Rather than questioning whether large language models are technically impressive, Zitron focused on something more fundamental: the unit economics.
According to him, companies developing frontier AI models continue spending extraordinary amounts on chips, data centres, electricity, and talent while struggling to demonstrate a path toward sustainable profitability. Subscription revenue and enterprise contracts, he argued, are not keeping pace with the capital required to build increasingly larger models.
He also suggested that AI leaders have entered an era where genuinely new hypergrowth markets are becoming harder to find.
In earlier decades, organizations could simply point to smartphones, cloud computing or social media as transformational growth engines. Currently, Zitron argues, AI has become the industry’s latest attempt to convince investors that another trillion-dollar opportunity is there even if commercial demand is yet to justify the spending.
Zitron’s Claims Trigger a Storm Across Social Media
Social media flooded with diverging views, with Zitron’s supporters on Reddit praising him for raising financial questions into a longstanding debate that revolves around benchmark scores and model capabilities. Many Reddit users also noted that skepticism about AI economics shouldn’t be confused with skepticism about AI technology itself.
One sentiment kept resurfacing throughout the discussion: AI can be genuinely useful while still being overpriced or massively overvalued as an investment.
Meanwhile, critics of Zitron argued that history rarely rewards people who dismiss transformative technologies too early.
They highlighted that cloud computing, smartphones, and the internet went through years of heavy spending before becoming profitable businesses. Taking this trend into consideration, today’s AI infrastructure investments could likely represent the early stages of a platform shift rather than evidence of a major AI bubble.
Note that while today’s AI models remain imperfect, their capabilities continue to improve rapidly. Historically, revolutionary technologies can coexist with investment bubbles. For instance, the dot-com bubble produced failures as well as some of the world’s most valuable companies. A similar outcome cannot be completely downplayed for artificial intelligence.
Zitron argued that many companies are caught in a costly race where each successive generation of AI models requires more computing power and higher investments simply to remain competitive. Overall, he believes that dynamic risks are creating an environment in which capex grows much faster than sustainable profits.
Leave a comment