The early innings of AI remind me a little of the Texas fracking boom of the early 2010s.
A lot of money is going out. Not a lot is coming in.
Frackers had to compete for leases, for equipment, and for drillers. Dozens of new companies were formed, and one even gave our family money for a lease that was never drilled, paying for my daughter’s college.
With AI it’s similar. Cloud Czars are bidding for Nvidia chips. Smaller companies are bidding for SuperMicro server racks. Cloud providers like Snowflake are bidding on any programmers who claim AI skills.
But the latest earnings reports show not a lot of money is coming in. Snowflake stock is down 20% today because what was a wildly successful business of renting cloud capacity now must be upgraded to serve AI applications that have yet to be written.
This has led a few people to turn bearish. But oil is being produced.
Real Progress
Klarna laid off 200 operators on its buy now, pay later play in 2022. Now most customers are being served by a single chatbot that works just as well .
Phone operations are low-hanging fruit for Generative AI. The input and output are the same. The work is clearly defined. When used for customer support, chatbots improved average productivity 14%. This jumped to 35% for new workers. This means more support delivered to more users, not fewer support positions.
AI is quickly turning the “Internet of Things” into the Machine Internet. Right now they’re just speeding repairs on complex machines, but once attached to masses of simple machines they create more efficient hospitals, universities, and stores. Drop the costs, connect them to utilities like water or traffic, and you have more efficient cities.
Despite scary headlines, the profitability of this low-hanging fruit is driving the AI revolution forward. While most of the money in AI is still going out, enough is coming in to make the effort worthwhile. And when the biggest investments start paying off, the world is going to change, mostly in a good way.
