Every few decades, a technology arrives that does not just improve how businesses operate. It reorganises them. Economists call these General Purpose Technologies. The steam engine, electricity, and the internet each met the criteria defined by Bresnahan and Trajtenberg in 1995: pervasiveness across sectors, continuous improvement over time, and the capacity to spawn complementary innovations. Generative AI has been classified as the latest, but with an adoption velocity that has no historical precedent.
Harvard and NBER research measured 39.4% adoption within 24 months of ChatGPT's public release, the fastest uptake of any General Purpose Technology in recorded history. The steam engine required 80 years to fully reshape manufacturing. Electricity took 40. The internet, approximately 20. GenAI is compressing that timeline into months, not decades. For two structural reasons.
The interface is natural language.
No specialized training is required to begin using the technology. Previous GPTs demanded new skills before they could be applied. GenAI meets people where they already are: in conversation.
The infrastructure already exists.
Previous GPTs required physical deployment: railways for steam, wiring for electricity, cables and servers for the internet. GenAI requires a browser. The delivery mechanism was built by the last GPT.
The Electricity Parallel and Its Implications
When factories first adopted electricity, most replaced their steam engines with electric motors and changed nothing else: same layout, same workflows, same management structure. The result was a modest efficiency gain. It took nearly a generation before manufacturers recognized that electricity enabled an entirely different factory design: smaller motors at each workstation, flexible floor plans, workflows impossible under a single-drive-shaft model. That second wave, the redesign wave, is where the transformative economic value was created. GenAI is currently in the equivalent of the bolt-on-a-motor phase. The redesign phase is beginning.
Virginia's 780,000 small and mid-size businesses sit at the beginning of this pattern. Most are in the bolt-on phase, encountering AI through their existing software without deliberate engagement. Some are beginning to experiment. A small number are already redesigning how they work around what the technology makes possible. The question this research explores is not whether Virginia businesses will encounter GenAI. They already have. It is whether they will develop the fluency to move from passive exposure to genuine competitive advantage, and what determines the difference between those that do and those that don't.