During his address at CES 2026, Daniel Newman, the CEO of the Futurum Group, firmly challenged the notion that the artificial intelligence sector is experiencing an unsustainable market bubble. Instead, he stressed that the crucial obstacle facing the industry is a pressing global deficit in compute power, essential for maintaining what he characterizes as a "multi-decade super-cycle" in AI development.
Newman articulated that concerns over a bubble stem from a misunderstanding of the current landscape. The surge in AI interest and investment is only the nascent phase of an extensive technological transformation. Central to this evolution is the emergence of "agentic" AI systems—advanced models capable of executing complex, autonomous operations beyond generating text or performing simple tasks. This transition is expected to drastically increase the demand for computational resources.
"We don’t possess enough computational infrastructure," Newman warned, using the metaphor of lacking sufficient turbines to meet the load. He highlighted that the current production and deployment of hardware cannot scale to satisfy today's demands, let alone those projected as AI applications become more sophisticated. His forecast anticipates a severe hardware shortage emerging within five to ten years as demand unfolds fully.
Looking ahead to 2026, Newman pinpointed this year as a pivotal moment when enterprise AI applications will become prominent and demonstrate tangible financial returns. He indicated that the industry is only utilizing a small fraction of the existing repositories of trained data, with much of it currently sequestered within proprietary corporate environments focusing on domains such as drug discovery, manufacturing optimization, and supply chain management.
This shift from predominantly consumer-oriented AI use cases, such as chatbots, towards enterprise-level deployment aligns with broader industry expectations that AI is entering a new phase. The "build phase," characterized by intense capital investment into training expansive AI models, is giving way to a "monetization phase," where real-world implementations will begin delivering clear productivity gains and improved margins for businesses leveraging AI inference capabilities.
To illustrate the scale and speed of AI operations, Newman cited the example of Alphabet Inc.’s Google and its Gemini AI model, which reportedly processes approximately 10 trillion tokens every day. This figure exemplifies the extensive computational requirements underpinning current AI workloads.
From a practical perspective, Newman also shared insights from his own firm’s experience, noting that AI has significantly accelerated market research processes. Tasks that traditionally required six months are now completed in two weeks, exemplifying efficiency improvements powered by AI technology.
Despite the skepticism expressed by some observers, Newman emphasized that the industry remains in the early stages of realizing AI’s full monetization potential and long-term durability. The structural challenges related to compute infrastructure must be addressed for AI’s impact to be sustained and scalable.
Key Points:
- Daniel Newman refutes the idea of an AI market bubble, highlighting a global shortage of computing resources as the true challenge.
- “Agentic” AI, capable of autonomous complex tasks, is expected to drive a steep rise in compute demand.
- 2026 is identified as the year enterprise AI will start producing measurable returns on investment, moving beyond consumer chatbot applications.
- Alphabet's Google Gemini model currently processes 10 trillion tokens daily, highlighting the massive computational scale involved.
Risks and Uncertainties:
- Insufficient hardware manufacturing capacity could bottleneck AI industry growth over the next decade.
- The currently underutilized trained data within proprietary corporate systems may not be fully leveraged in expected timelines.
- Skepticism persists regarding the timing and scale of AI’s transition from development to profitable commercialization.
- Supply chain or technology constraints could hamper enterprise AI deployments planned for the near future.