AI adoption follows the J-curve path of general-purpose tech, like early US factory electrification, requiring years of investment before noticeable ROI gains
Exponential View
Context & Ripple Effects
Related coverage places this argument inside an AI investment cycle marked by unusually heavy infrastructure spending and still-thin margins: reported AI sales have exceeded estimated data-center and chip depreciation, but only narrowly.
The broader record is mixed rather than uniformly pessimistic. AI-exposed sectors have coincided with stronger productivity growth and weaker entry-level hiring, while rapidly improving models are changing workplace use and provoking policy and market adjustments.
First-order effects
Organizations deploying AI must treat integration, workflow redesign, data preparation, and complementary infrastructure as upfront costs rather than expect prompt, easily measured returns.
AI suppliers and large infrastructure buyers remain exposed to a timing mismatch: investment and depreciation are immediate, while customer productivity gains—and thus durable willingness to pay—take longer to establish.
Second-order effects
Buyers are likely to favor narrowly measurable deployments and tighter procurement scrutiny until broad workflow gains appear, raising pressure on vendors to demonstrate outcomes rather than model capability alone.
Thin economics during the buildout intensify competition across the AI stack, especially for providers whose revenue must support rapidly depreciating compute and data-center assets.
Third-order effects
If the J-curve pattern persists, AI advantage will accrue less to firms that merely purchase models than to those able to reorganize processes and sustain complementary investment through the low-return phase.
The AI cycle may increasingly resemble other capital-intensive general-purpose technology transitions: early spending concentrates among well-capitalized firms, while economy-wide productivity effects arrive unevenly and with labor-market disruption.
The trend: AI is moving from a rapid capability race into a slower diffusion phase in which organizational adaptation and capital discipline determine whether infrastructure spending becomes durable productivity growth.
GPTs are GPTs, and deploying them in a way that maximizes productivity growth is highly non-trivial This is the exact idea behind the FDE team at OpenAI - we aim to drive maximum possible returns for our clients by reimagining workflows from the ground up with agents in mind
Invention and diffusion are separate concepts. We will have 10-20 years of work to realize the benefits of the AI that's already here. It won't be overnight. Spending a bunch of money on tokens does not a process re-engineer.
“Erik Brynjolfsson calls this the productivity J-curve: general-purpose technologies are a drag in their early years because firms have to make complementary intangible investments before the gains materialize.” [image]
There is no other industry, surely, where this mealy mouthed paragraph, void of any real measure, is presented as proof of ‘value’ and/or success? [embedded post]
Meanwhile, tensor processors are on a 3 year depreciation schedule and are nearly 40% of data center costs. Tick tock! — My heart will burst with joy if adoption takes so long, the worst humans ever to live lose absolutely existential amounts of money, and the infrastructure h…