Folks worried about an AI bubble and doubting if we're gonna see a repeat of the internet bubble burst just don’t get a few key facts:
1. Top AI firms are sitting at a PE of only 20-30x, which is actually pretty chill, even low. Back during the bubble, Cisco was rocking a PE of 150x—and they weren't even selling scarce computing power, just routers and cables.
2. Revenue for leading AI companies is hitting new highs, and the demand for AI from humans is only gonna skyrocket. We're just getting started, and the adoption rate in boosting productivity across industries is still super low—can you even imagine going back to an AI-less world? Nah, you can’t.
3. AI has only been around for a few years, the hardware rollout isn’t even complete, and we're still on the verge of software explosions, application breakthroughs, and productivity booms. The future has robots, space exploration, healthcare, and quantum computing all tied to AI development.
Back in 2000, the bubble's typical issue was that a ton of companies couldn't turn a profit, and telecom infrastructure was way overbuilt, with fiber optics, switches, and network capacity outpacing actual demand.
But today, it’s the complete opposite: computing power, storage, bandwidth, inference costs, model capacities, and application scenarios are all rapidly expanding. It’s not that there’s a lack of demand; it’s that the infrastructure isn’t cheap enough, plentiful enough, or good enough.
Blindly comparing bubbles, or just turning bearish because prices have shot up, shows a lack of deep research—it's toxic thinking straight outta a daydream.
1. Top AI firms are sitting at a PE of only 20-30x, which is actually pretty chill, even low. Back during the bubble, Cisco was rocking a PE of 150x—and they weren't even selling scarce computing power, just routers and cables.
2. Revenue for leading AI companies is hitting new highs, and the demand for AI from humans is only gonna skyrocket. We're just getting started, and the adoption rate in boosting productivity across industries is still super low—can you even imagine going back to an AI-less world? Nah, you can’t.
3. AI has only been around for a few years, the hardware rollout isn’t even complete, and we're still on the verge of software explosions, application breakthroughs, and productivity booms. The future has robots, space exploration, healthcare, and quantum computing all tied to AI development.
Back in 2000, the bubble's typical issue was that a ton of companies couldn't turn a profit, and telecom infrastructure was way overbuilt, with fiber optics, switches, and network capacity outpacing actual demand.
But today, it’s the complete opposite: computing power, storage, bandwidth, inference costs, model capacities, and application scenarios are all rapidly expanding. It’s not that there’s a lack of demand; it’s that the infrastructure isn’t cheap enough, plentiful enough, or good enough.
Blindly comparing bubbles, or just turning bearish because prices have shot up, shows a lack of deep research—it's toxic thinking straight outta a daydream.