AI spreads fast because there is nothing to install
Business software had to be wired into old systems like a dishwasher, while AI works the moment you open a window.
Invest Like the Best with Patrick O'Shaughnessy · Why the AI Boom Is Just Getting StartedTo understand how fast artificial intelligence can spread, it helps to think about a radio and a dishwasher.
That comparison comes from Alex, an investor at Whale Rock Capital, whose firm hired a researcher who once worked with the late Harvard professor Clayton Christensen to chart the great adoption booms of the past 100 years. The radio was one of the fastest ever, reaching something like full ownership in American homes in about seven years. The dishwasher was slow and flat by contrast, because it had to be plumbed into the back of the house.
The difference is what has to happen before first use. A radio just needed to be bought, plugged in, and switched on. A dishwasher had to fit the kitchen, connect to water and drains, and work with the way a household already ran. Consumer products that work on their own tend to spread quickly, while business tools that must connect to existing systems take much longer.
That plumbing problem is why business software moved slowly. Alex puts cloud computing and its subscription software in the dishwasher camp. Big companies were worried that keeping data outside their own walls was unsafe, and information technology departments called it risky. Adoption picked up after large and cautious organizations, including the CIA and Capital One, moved data to the cloud, and Capital One's technology chief said it was actually more secure there. Even then, growth ran at a strong but ordinary pace, roughly 30 to 50 percent a year, because every customer still had to connect the new system to the old one.
AI is different because there is often nothing to install. For a consumer, and even for many workers, you just open a web browser and it is there. No pipes, no rewiring, and no long integration project stand between curiosity and use.
That is why Alex does not describe AI as a normal S-shaped adoption curve, which starts slowly, rises, and then levels off. He calls it a backward L, a line that shoots almost straight up. The takeaway is simple: when a technology needs plumbing, watch the plumbers, but when it needs only a browser, expect speed.
Questions people ask
What is an S-curve of adoption?
It is the S-shaped path many products follow, with slow uptake at first, fast growth in the middle, and a slowdown as most potential buyers already own it. Radios traced a steep S, while dishwashers traced a much flatter one. Investors use the shape to judge how big and how fast a market may become.
Why was cloud computing slower to catch on?
Cloud computing meant storing company data and programs in large outside data centers rather than on computers a company owned itself. Many executives feared that would be less secure and harder to connect to older systems. Adoption accelerated once trusted users showed it worked and security worries eased, but each company still needed an installation project.
Does AI really require no setup for businesses?
No, not always. Trying a chatbot takes only a browser, but using AI across a large company can still raise security, privacy, and training questions. The point is that the first step is unusually easy, so experimentation can spread before the harder work of connecting AI to company records and rules.