After seeing the recent divergent paths in Oracle’s and Salesforce’s stock performance, I decided to do a little research into why. It appears that Oracle is benefiting from investing in AI backend and infrastructure, while Salesforce has focused on AI applications. This got me thinking about the disconnect in the AI industry. I keep reading headlines about the growth in AI backend and infrastructure, but then also read headlines that indicate the AI front end seems to be hitting a wall in terms of realized ROI and effective adoption.
Curious about this seeming disconnect, I looked for historical instances of this pattern of infrastructure outpacing application.
History offers a few parallels:
Railroads: Tracks were laid across empty plains before the towns and farms appeared.
Electrical Grids: Power plants were built before a wave of household appliances was invented to consume the electricity.
The “Dark Fiber” Bubble: Perhaps the closest comparison to today’s AI landscape is the “dark fiber” glut of the dot-com era.
In the late 90s, telecom companies laid a massive surplus of fiber optic cable, betting on a future of unlimited internet demand. The backend infrastructure far outpaced the actual frontend applications of the day.
Soon after the market crashed.
However, that overbuilt, “dark” infrastructure didn’t go to waste in the end. It became the cheap and abundant foundation that enabled the next wave of innovation: YouTube, cloud computing, and the entire streaming economy.
Today, there are immense investments in AI compute capacity and foundational models, the new “fiber.” The challenge is that we’re still searching for the killer apps to truly “light it all up” and justify the investment.
The question is, are we just in a temporary lag before an explosion of innovation, similar to the post-dot-com era? Or is the challenge of creating valuable applications more fundamental this time?
What do you think?
Recent Comments