I keep coming back to one question when I look at AI today. How long can this level of spending continue?
The numbers are staggering. Billions poured into training single models, hundreds of billions committed to infrastructure. And most of the major players are still far from profitability.
But this isn’t irrational spending. It’s a calculated race. Every company knows that being second or third in AI may not be enough. The upside is so massive that it justifies the burn, at least for now.
Why Profitability Isn’t the Priority Yet
At first glance, the business model looks shaky. Subscription plans seem simple, but they hide a deeper issue.
Heavy users cost significantly more to serve than casual ones. In some cases, the gap is extreme. That means these subscriptions aren’t purely about revenue. They’re about retention, data, and positioning.
Right now, the goal isn’t efficiency. It’s dominance. Build the best system, attract the most users, and figure out monetization later.
It’s a familiar pattern in tech, just at a much larger scale.
The Overlooked Trend: Costs Are Falling Fast
What is often overlooked in this conversation is how quickly things are improving behind the scenes.
The cost to achieve the same level of performance has dropped dramatically in just a few years. What was once expensive is becoming accessible at scale.
This creates an interesting dynamic. While cutting-edge models are incredibly costly, older capabilities are getting cheaper at an equally aggressive pace.
So, even if progress slows at the frontier, the value of what already exists continues to expand.
Will the AI Bubble Burst?
I don’t see a sudden collapse where AI disappears from everyday use. That seems unlikely.
A more realistic outcome is a slowdown. At some point, the cost of pushing the frontier further may outweigh the immediate returns. When that happens, investment could tighten, and expectations may reset.
But that doesn’t erase the utility. Just like previous tech cycles, the underlying value remains even if the hype fades.
A Future Controlled by a Few
One consequence feels almost inevitable.
If costs keep rising at the top end, only a handful of companies will be able to compete. Training and maintaining advanced models may become concentrated among two or three dominant players.
Others will build on top of them, but won’t replicate the same infrastructure.
For me, the real shift isn’t whether AI survives. It’s how the industry consolidates and evolves from here.
Because even if growth slows, the role of AI in everyday work is already too embedded to reverse.
Follow Us on:
Clutch
Goodfirms
Linkedin
Instagram
Facebook
Youtube
