For three years, one of the biggest questions surrounding AI has been simple: where are the productivity gains?
Companies have poured hundreds of billions of dollars into AI infrastructure, tools and talent. Yet national productivity statistics have remained stubbornly less dramatic than the technology's advocates predicted. That gap has fueled a growing chorus of skeptics wondering whether AI hype has outrun AI's measurable economic impact.
A newer economic framework offers a different explanation, according to reporting from South Korean outlet Maeil Business Newspaper.
The idea is called the "productivity J-curve." It suggests that the disappointing productivity numbers may be a normal part of introducing a transformative technology rather than evidence that the technology isn't working.
Why Productivity Can Dip Before It Rises
The J-curve concept isn't new to economics. It describes situations in which the benefits of a major change take time to appear in measured results.
Applied to AI, the logic is straightforward.
Businesses spend enormous sums on chips, data centers, model training, software and employee retraining. Much of that spending initially shows up as investment and cost rather than measurable output.
The productivity gains come later, once companies redesign workflows, retrain employees and change how work is actually performed around the new technology.
That transition can take years.
Economists have pointed to electricity and computers as historical parallels. Both technologies required substantial investment and organizational change before their effects became clearly visible in productivity statistics.
The argument now gaining attention is that AI may have reached, or passed, the deepest part of that initial productivity dip.
What "Past the bottom" Actually Means
This is not a claim that AI has already delivered its promised economic windfall.
It is a claim about direction.
Under the J-curve framework, the least productive phase of AI investment may occur before the technology becomes deeply integrated into everyday business processes. Infrastructure buildout, foundational model development and early trial-and-error deployments represent much of the initial cost.
The potential payoff comes later through workflow redesign, workforce adaptation and increasingly capable tools that become cheaper to deploy.
If that interpretation is correct, the absence of dramatic productivity gains so far does not necessarily mean AI has failed.
It could mean the economy is still working through the expensive transition period.
That offers a meaningfully different explanation from the more pessimistic narrative that gained momentum through 2024 and 2025, when several high-profile studies questioned whether generative AI was producing measurable gains inside large organizations.
Why The Timing Matters
If the J-curve framework holds, the next two to three years could be particularly important.
That is the period in which productivity data may begin to provide stronger evidence for either side of the argument.
For investors, the distinction matters.
Companies including Microsoft, Google, Amazon and Nvidia have committed enormous amounts of capital to AI infrastructure partly on the expectation that productivity gains will eventually translate into broader economic demand for AI products and services.
If productivity remains weak, that investment thesis becomes harder to defend.
If productivity begins accelerating, it could validate years of spending that critics have described as excessive.
The important point is that the J-curve is a framework for interpreting what happens next, not proof that the AI productivity boom is inevitable.
The data still has to turn.
What This Means for Miami
Miami's role in this story isn't primarily as a research hub. It is as an adoption market.
The city has positioned itself as a magnet for AI-forward companies, fintech firms and startups betting that AI will reshape how businesses operate.
If the productivity J-curve framework proves accurate, the potential payoff will favor companies willing to redesign workflows rather than simply add AI tools on top of existing processes.
That is particularly relevant to Miami's fintech, healthcare and real estate sectors, where businesses are increasingly experimenting with AI but may not yet be seeing the full economic benefit.
For local investors and venture firms backing AI startups, the framework offers another way to think about the current cycle.
Weak productivity numbers do not necessarily prove that AI investment is failing.
They may simply indicate where the economy sits on the curve.
The next few years of productivity data will determine whether that optimism is justified.
