Corporate demand for AI computing resources continues to far exceed supply, even as chief financial officers begin applying stricter scrutiny to AI budgets. CNBC reported on December 12, based on interviews with former Intel CEO Pat Gelsinger and executives at AI infrastructure companies, that the shift represents a move toward investment efficiency rather than a pullback.
Gelsinger, now CEO of venture capital firm Playground Global, described AI demand as "practically close to unlimited" and said the only constraint is power supply. "The economic value AI will create is close to infinite across nearly every industry," he said.
Mark Boroditsky, chief revenue officer at Nevious, a neocloud company that builds Nvidia GPU-based data centers and rents them to AI startups, said the demand the company sees "far exceeds what we can handle." He added: "This situation has already continued for quite some time."
Oversupply fears meet industry pushback
Recent moves by Meta and xAI to rent out surplus AI computing capacity stirred concerns that AI infrastructure might be oversupplied. The share prices of AI semiconductor companies such as Samsung Electronics and SK hynix also wobbled on those fears.
Industry executives dismissed the concern as a misreading of isolated cases. Andrew Feldman, CEO of AI chip company Cerebras Systems, called the Meta and xAI situations "exceptional" and said data centers and compute infrastructure remain in short supply across the broader industry. Park Seong-yoon, CEO of AI semiconductor startup Rebellions - which counts Samsung Electronics and SK hynix as investors - said the renting of excess capacity "does not mean hyperscalers have overinvested."
Michael Hulsen, CEO of optical communications equipment maker Lumentum, reinforced the point: "Most of what we will produce over the next five years has already been sold." The company is adding capacity to meet demand projected five years out.
A new phase: maximizing return on AI investment
While demand remains strong, the approach to AI spending is changing. Boroditsky said CFOs have started to scrutinize AI budgets more strictly, but framed it as a shift to a "value maximizing" phase that prioritizes investment efficiency. For executives building AI strategy, this signals a move from broad adoption to measured deployment - a topic covered in resources on AI for Executives & Strategy.
Feldman predicted that companies will increasingly match AI models to the complexity of the task. "State-of-the-art AI models will be used for complex tasks, and lightweight models will handle simple work," he said. "It's the same principle as not using a large vehicle every time you go around the neighborhood."
Why this matters for executives and strategy leaders
The message from the infrastructure layer is clear: AI demand is not softening. The constraint is supply, particularly energy capacity. But the era of unchecked AI spending is giving way to a more disciplined approach. CFOs are demanding ROI, not retreating. For strategy leaders, the practical implication is twofold: secure compute capacity now while supply remains tight, and build internal frameworks for matching AI model cost to task complexity. Organizations that treat AI spending as a portfolio to be optimized - rather than a budget line to grow indefinitely - will be better positioned as the market matures.
Your membership also unlocks:
