Introduction

AI spending is rising fast, but real money from users is missing. We now face a $600B gap between cost and value.

What's the problem it solves?

Tech giants are spending huge amounts to build AI data centers, but the money they earn from users isn’t keeping up. This article explores whether these investments are truly smart, or if we’re heading into a big AI money trap.

Quick Summary

In late 2023, AI companies were already spending more than they were making. Fast forward to mid-2024, and the spending has grown even faster, turning a $200 billion problem into a $600 billion one. Nvidia is now the world’s most valuable company because everyone is buying its AI chips. But there's a key question: is this growth real, or just hype?

While companies like OpenAI are making money, most others are not. Many businesses are hoarding GPUs like they’re gold, but not all of them are using them well. There’s also little control over prices because too many companies are joining the AI race. Plus, older chips lose value quickly as newer ones come out. This could hurt investors the most. On the bright side, as prices fall and learning increases, smart builders will have a better chance to succeed.

Key Takeaways from the article

• AI spending has jumped from $200B to $600B, but most companies aren’t earning enough to match that cost.
• GPU supply issues are over, but companies are still stockpiling them heavily.
• A few companies like OpenAI are making real money; most others are not.
• AI infrastructure is getting built fast, but without pricing power, it may not pay off.
• Older AI chips lose value quickly as new ones (like Nvidia’s B100) arrive.
• Many investors may lose money, but AI startups might benefit from cheaper tools.
• Believing AI will instantly make everyone rich is dangerous thinking.
• Building solid products that truly help people is still the best long-term path.

‍