Ed Zitron critiques the AI investment narrative by highlighting that major tech companies’ cloud revenues heavily depend on just two unprofitable AI firms, OpenAI and Anthropic, creating a fragile and unsustainable financial cycle. He warns that this concentration, combined with massive infrastructure costs and delayed IPOs, poses significant risks to the broader AI ecosystem’s growth and profitability.
Ed Zitron provides a critical perspective on the current AI investment narrative, highlighting that major tech companies like Microsoft, Google, and Amazon have significant portions of their cloud revenues tied to just two AI companies: OpenAI and Anthropic. UBS estimates that nearly half of Google Cloud’s revenue next year will come from these two firms, while Barclays reports a growing share for Microsoft as well. Zitron argues that this concentration is misleading investors into believing that AI revenue streams are diverse and sustainable, when in reality, these two companies are unprofitable and heavily reliant on continuous capital infusions.
Zitron emphasizes that OpenAI and Anthropic are essentially burning through vast amounts of cash without generating sustainable profits. For example, OpenAI reportedly lost $20.9 billion in 2025 alone, with much of its revenue coming from questionable sources like SoftBank’s Crystal Intelligence program. Despite their massive funding rounds and infrastructure support from hyperscalers, these AI firms remain financially unstable. The infrastructure costs are enormous, with Microsoft and Google having invested tens of billions in data centers and compute resources that only a few companies can afford to utilize at scale.
The concentration of AI revenue in just two companies raises concerns about the broader AI ecosystem’s health and sustainability. Zitron points out that the data center capacity being built globally requires trillions in revenue to be viable, but with only a handful of customers like OpenAI and Anthropic, this demand is not being met. This creates a circular financing problem where hyperscalers invest heavily in infrastructure primarily to support these two AI firms, which in turn depend on the hyperscalers for their compute needs, creating a fragile and potentially unsustainable cycle.
Looking ahead, Zitron identifies the delayed IPO of OpenAI as a critical flashpoint that could expose vulnerabilities in the AI market. Both OpenAI and Anthropic need continuous capital injections to survive, and any disruption in funding could trigger a domino effect impacting the hyperscalers and the broader AI investment landscape. Furthermore, the slow pace of data center construction and the enormous capital requirements mean that the AI industry’s growth is constrained by physical and financial realities, challenging the optimistic narratives around AI-driven productivity gains and returns on investment.
Finally, Zitron warns that the current AI hype may resemble past financial scandals where smart executives created complex but ultimately unsustainable business models. While the major tech companies likely conducted due diligence, their heavy reliance on just two AI customers and the circular nature of their investments raise red flags. The future of AI profitability hinges on whether OpenAI and Anthropic can scale sufficiently to justify the massive infrastructure investments, a scenario that remains uncertain given their current financial trajectories and the competitive pressures from emerging players like China.