Nvidia isn’t Enron. I think the AI boom has barely started.
I keep seeing the Nvidia/Enron comparison after Michael Burry’s comments. I just don’t see it.
There are fair questions about Nvidia’s valuation and whether we’re spending too much on AI infrastructure. But Enron was fundamentally about hidden debt, questionable accounting and profits that weren’t what investors thought they were.
Nvidia is selling a very real product to customers spending enormous amounts because they need compute.
The bigger question is: **are we actually building too many data centres?**
I think there’s a chance we’re underestimating future demand.
The internet made information available to everyone. AI is starting to make **intelligence available to everyone.**
Think about every student having a personal tutor. Every small business having access to programmers, designers, analysts and accountants. Millions of people becoming 2x, 5x or potentially 10x more productive.
And humans won’t be the only users.
AI agents will work continuously. Robots, cars, factories and software will consume AI without a human sitting there typing prompts.
There are 8 billion humans, but eventually there could be many times that number of AI agents doing work.
People also say AI will become more efficient and therefore need less compute.
Maybe. But bandwidth became cheaper and we didn’t use less internet — we invented Netflix, YouTube, cloud computing and 4K streaming.
If AI becomes 10x cheaper, we may find 100x more uses for it.
That’s why Cisco is a much better bear comparison than Enron. Cisco was a real company selling real infrastructure into a real revolution. The question was whether too much was built too quickly.
That’s a legitimate debate with Nvidia.
But my bet is that **AI is becoming a utility — intelligence on demand.**
Electricity gave us energy on demand.
The internet gave us information on demand.
AI could give us intelligence on demand.
And when humans **and machines** start consuming that intelligence continuously, today’s data-centre buildout may not look so crazy.
**Maybe we’re not near the end of the AI boom. Maybe we’re still at the beginning.**