Anticipating the Gain
When a company adopts AI, benefits materialize after a brief period where results are flat or might even experience a dip. Economists call this pattern the productivity J-curve. Understanding it changes how you budget for AI and how you judge the first year.
Erik Brynjolfsson, who directs Stanford University’s Digital Economy Lab, developed the idea with two colleagues.¹ Buying the tool is only the first step. The value comes when people learn new skills and the company redesigns how the work gets done. That effort takes time and money before it produces more output. Brynjolfsson says it can cost up to ten times what the technology does, and most of it is never measured.² It appears as people being busy with something other than their regular work.
This has happened before. Companies invested heavily in personal computers in the 1980s. Getting the benefit required further investment in software, retraining and reorganizing, and measurable returns did not appear until the late 1990s.³
There are signs the curve is turning upward. Brynjolfsson calculates that productivity in the United States grew roughly 2.7 percent in 2025, nearly double the 1.4 percent average of the previous decade. He reads that as earlier efforts beginning to pay off, and he cautions that several more periods of growth are needed to confirm the trend.⁴
For a business owner, the J-curve leads to three decisions. First, budget for the change as well as the tool, because training people and redesigning a process are part of the cost of AI. Second, do not judge the early months by profit. Revenue and profit are the last numbers to move. Track what moves first, such as the time saved per person and whether AI is being used on work that matters to the business. Third, shorten the dip by concentrating on one important process, and avoid spreading the effort across many.⁵
The J-curve is an explanation, and it should not become an excuse. A dip is only an investment if the redesign work is under way. If nothing has changed about how the work gets done, flat results are simply flat results.
In “When Will We See Benefits of AI?” I wrote that every AI initiative needs a business case that estimates when the benefits will outweigh the costs. The J-curve is the reason that date is usually later than people expect.
If your AI results look flat today, ask one question. What have we changed about how the work gets done? If the answer is very little, that is the place to start. If the answer is a great deal, you are probably in the dip, and the gain is ahead of you.
Sources
¹ Erik Brynjolfsson, Daniel Rock and Chad Syverson, “The Productivity J-Curve: How Intangibles Complement General Purpose Technologies,” American Economic Journal: Macroeconomics, 2021.
² Erik Brynjolfsson, interviewed by Nicholas Thompson, “What AI Is Really Doing to Jobs Right Now,” The Atlantic, video, May 27, 2026.
³ Neale Mahoney, Erika McEntarfer and Karsen Wahal, “What is really happening to jobs? Separating AI hype from reality,” Stanford Institute for Economic Policy Research, policy brief, July 2026.
⁴ Erik Brynjolfsson’s analysis in the Financial Times, as reported by Jason Ma, “One of Stanford’s original AI gurus says productivity liftoff has begun after doubling in 2025 amid transition to ‘harvest phase’ along J-curve,” Fortune, February 15, 2026.
⁵ Erik Brynjolfsson, Andrew McAfee and James Milin, “The ROI Gap: What 2026 Data Reveals About Enterprise AI Returns,” video, August 18, 2026. The three are co-founders of Workhelix, an AI consulting firm.