The AI industry’s growth-at-all-costs mindset is facing a reality check as even its biggest players confront the staggering expense of powering the revolution.
OpenAI has missed key revenue and user growth targets, according to a report on April 27, 2026, prompting internal debate over the wisdom of its massive data-center spending and signaling a potential valuation crunch ahead of its highly anticipated IPO.
The competitive landscape is fierce, with rivals like Anthropic growing at a pace OpenAI's own Chief Revenue Officer, Denise Dresser, called "inflated" in a recent internal memo. "They use accounting treatment that makes revenue look bigger than it is," Dresser wrote, claiming Anthropic's $30 billion run rate is overstated by roughly $8 billion.
The internal friction at OpenAI centers on massive capital expenditures for data centers in the face of slowing growth. This spending boom is industry-wide, with Meta Platforms recently launching a program to train fiber technicians to staff its own data center buildout. The scale of these investments is immense; Amazon's CEO Andy Jassy recently noted that if its in-house silicon business sold chips directly, it would represent a nearly $50 billion annual revenue run rate, putting it ahead of AMD.
This news could temper enthusiasm for OpenAI's eventual IPO and create negative sentiment for the broader AI sector. For investors, the key question is where value is truly being created: with the high-profile model makers like OpenAI and Anthropic, or the infrastructure providers like Amazon, Google, and Intel who supply the costly computational power.
The AI boom is forcing a strategic re-evaluation across big tech. While OpenAI and Anthropic capture headlines, the foundational layer of semiconductors and cloud infrastructure is where billions are being spent. Amazon's AWS is a prime example, with its custom Graviton and Trainium chips attracting major AI players. Meta recently agreed to use Graviton CPUs, while both OpenAI and Anthropic have committed to using Trainium chips for training and inference.
Google is also a major contender, debuting its new TPU 8t and 8i processors to compete directly with Nvidia and AMD. The company has secured multi-billion dollar deals to provide its TPU capacity to both Anthropic and Meta, highlighting the intense demand for specialized AI hardware.
The rivalry between the leading AI labs is escalating beyond technology into public criticism. Dresser's memo attacking Anthropic's accounting and "single-product company" focus reveals the high stakes of what she termed a "platform war." Anthropic, for its part, recently released its Claude Opus 4.7 model, while carefully limiting access to its more powerful, and potentially hazardous, Mythos model.
This backdrop of intense competition and massive spending commitments makes OpenAI's reported growth slowdown particularly significant. It suggests that even with cutting-edge technology, the path to profitability is fraught with challenges, and the enormous cost of staying at the forefront of AI development may not be sustainable without consistent, exponential growth.
This article is for informational purposes only and does not constitute investment advice.
