Real AI Crisis Is Compute Shortage, Not Bubble, Says Daniel Newman: 'Those Calling For A Bubble Don't Understand What's Happening'

Speaking at the CES 2026, Futurum Group CEO Daniel Newman dismissed growing concerns of an artificial intelligence (AI) market bubble, arguing instead that the industry faces a far more tangible crisis: a severe global shortage of compute power necessary to sustain a “multi-decade super-cycle.”

The Infrastructure Gap

“Those calling for a bubble really don’t understand what’s happening,” Newman said in an interview clip shared on his X account. He posits that the current market frenzy is merely the preliminary stage of a long-term technological shift.

According to Newman, the rapid rise of “agentic” AI systems capable of executing complex autonomous actions rather than just generating text—will soon spike compute intensity to unsustainable levels.

“We don’t have enough turbines. We can’t build what the demand is for what people are doing today,” Newman warned. He predicts that as demand fully activates over the next 5 to 10 years, the technology sector will face a critical hardware shortage.

2026: The Year Of Enterprise ROI

Newman highlighted 2026 as the inflection point where “enterprise AI will show its face,” moving the narrative beyond consumer chatbots to substantial corporate returns on investment (ROI).

He noted that the industry is currently utilizing only a small percentage of available trained data, with the vast majority still locked inside proprietary corporate systems used for drug discovery, manufacturing, and supply chain optimization.

This shift aligns with broader market sentiment that the industry is transitioning from the capital-intensive “build phase” of training models to the “monetization phase” of inference, where AI begins delivering measurable margin improvements and productivity gains.

Efficiency At Scale

To illustrate the sheer scale of current activity, Newman pointed to a staggering metric: Alphabet Inc.-owned (NASDAQ:GOOG) (NASDAQ:GOOGL) Google’s Gemini AI model is now generating 10 trillion tokens daily.

On a practical level, he cited his own firm's operational efficiency, noting that AI has compressed complex market research workflows that previously took six months into just two weeks.

“We are very early,” Newman concluded, emphasizing that despite the skeptics, the era of significant AI monetization and durability has only just begun.

Here’s a list of AI-linked ETFs for investors to consider:

Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.

Photo courtesy: Shutterstock