The debate over AI pacing is being framed in Washington and Silicon Valley as a race issue. But the numbers in this debate point to something more basic: most companies are not ready to absorb frontier AI at scale.
In a commentary published by Fortune, Jeffrey Sonnenfeld, David Siegel, and Stephen Henriques argue that Corporate America is already years behind the AI frontier. They say the real question is not whether the labs can keep pushing capability, but whether businesses can actually adopt what already exists.
Their case rests on enterprise constraints that are hard to ignore. They point to fragmented data silos, legacy ERPs, strict compliance regimes, and basic data hygiene as barriers to true economic absorption. They also cite a set of company-level figures: more than two-thirds of high-performing companies identify data as the primary barrier to implementing AI; only 7% describe their data as 'completely ready' for AI; fewer than a quarter have a data strategy at all; and 63% either lack AI-suitable data management or are unsure whether they have it.
The authors also argue that technical availability is not the same as economic transformation. They cite McKinsey Senior Partner Asutosh Padhi saying that general-purpose technologies have historically required decades to reorganize workflows and generate broad productivity gains. They add that electricity took 75 years to lift productivity economy-wide, computers required 50 years, and the internet and mobile devices demanded 25.
The pace of corporate uptake, they say, has been fast in rhetoric and slow in practice. Since ChatGPT was released in 2022, executive suites have moved with 'unprecedented urgency,' but the authors say many daily enterprise workflows do not require a frontier system at all. They also say that 'tokenmaxxing' has produced budget shocks and that most firms are concentrating on lower-risk automation tasks.
The piece acknowledges the counterargument too. The authors say pacing skeptics are not frivolous and note concerns that slowing development could cede the U.S. lead to China. They also say warnings about catastrophic or existential risk can no longer be dismissed outright.
Still, their bottom line is that the economy is not yet at the frontier people assume. If the bottleneck is adoption, not invention, then the market has already voted on the near-term value of AI: useful, yes, but still trapped inside the limits of enterprise infrastructure.
That matters for capital allocation. The hype cycle rewards talk of transformation. Free enterprise rewards what companies can actually deploy, govern, and scale. If the data is not ready, the productivity story is not ready either.
For investors and managers, the lesson is straightforward: the frontier is not the same thing as broad-based earnings power. And until businesses fix the plumbing, the administrative state should not be the only one accused of moving too slowly.
