For a start this isn’t a political or moral piece on AI at all. Nor is it saying sell cause the bubble is about to burst. What I’m aiming for is an honest piece that makes everyone who reads it aware of the pros and cons of where we currently sit and how to keep on your toes trading it.
AI has once again been in focus this week. Last week was down to earnings, this week the midterms. The question is : is this 2000 all over again?
I suppose the honest answer is yes and no. It rhymes, but it doesn’t repeat.
Why it feels like 2000
The Nasdaq peaked at 5,048 on 10 March 2000. Two and a half years later it had lost roughly 78% of its value. People who bought the story late didn’t get a gentle pullback.
Some of what we’re seeing now will feel familiar to anyone who was around then:
Concentration. The top ten stocks make up about 36% of the S&P 500, against 27% in March 2000. A handful of names now carry the index.
The spending. Amazon, Alphabet, Microsoft and Meta have guided to roughly $730 billion of capex for 2026, up 78% from about $410 billion in 2025.
The cash gap. In the latest quarter, capex exceeded operating cash flow at Alphabet, Amazon and Meta, and only Microsoft was the exception.
Why it’s different
The biggest difference is who is doing the spending, and what they earn.
In March 2000, Barron’s found that 74% of 207 internet companies had negative cash flow. Today’s leaders are profit machines. Only around 20% of tech companies are unprofitable now.
Valuations are also lower, although not low. The Nasdaq-100 traded at about 60 times forward earnings in March 2000. The S&P 500 sits at roughly 22 times forward earnings now, against 27 then. Expensive, yes. Dot-com expensive, no.
What about jobs?
This is the part I find most interesting, because the market and the labour market are telling different stories. It’s also the part that, when you look at where revenues in AI come from, does scare me.
ES is near 8,000, and the headline unemployment rate was 4.2% in September 2026. That doesn’t look like a market worried about work.
But look at the bottom rung. Recent graduates had unemployment of about 5.6% in the second quarter, with 42% underemployed in jobs that don’t need a degree. Stanford, using payroll records, found 22-to-25-year-olds in AI-exposed jobs have 19 percent fewer jobs than they would have if they’d tracked their less-exposed peers.
There are surveys hitting mainstream media citing jobs that won’t exist in 10 years due to AI. Perhaps there have always been jobs throughout history that cease to exist but somehow these jobs lost feel different. I think it’s because the job still needs doing but just fulfilled by AI. Whereas, I don’t know, a steam train driver losing his job is down to the function of steam trains ceasing to exist. Feels better, though not for the driver, obviously!
Is that proof AI is taking jobs? I think it is.
AI is quietly changing who gets in. That’s a medium-term issue for consumer demand, and something to keep watching. Because it’s important and does bleed into revenues.
Revenues
AI companies run on razor thin margins. It is mostly consumer led. Subscriptions and tiered access. Software access from other vendors. Hardware upgrades for the likes of NVIDIA.
The challenge is this. If jobs are lost or at least unhired, consumers have less to spend. I’m sure we all hop from ChatGPT to Claude to Gemini without looking to spend. And they rely on us spending.
Every search you do for free costs the AI companies a fortune. Immense server power and electricity. It’s called INFERENCE cost. Every search is unique. Every interaction costs money. It’s not like software that once written can be rolled out at no cost. And don’t forget the “training” of said AI. Hundreds of millions spent before we even catch a glimpse of it.
This is why I think the initial sell-off on the midterms was correct. Because if, as expected, the Democrats slow the roll of AI with red tape and stricter environmental procedures is could be big trouble.
The IPO wave
There’s another sign of how big this theme has become: the listings.
The record is currently SpaceX. It raised $75 billion. And it may not hold the title for long. Reuters has reported that Anthropic is seeking to raise as much as $100 billion. The listing could easily value the company at $2 trillion.
Why I think it’s a forever trade
I think AI is infrastructure, like electricity or the internet. We won’t go back. The dot-com crash didn’t kill the internet. It killed the prices people paid for it, and the companies that survived went on to define the next twenty-five years.
But “forever” and “straight line” are very different things. A theme can be permanent while an entry is terrible. Someone who bought the Nasdaq in March 2000 was right about the internet and wrong about the price, and it took them about fifteen years to get back to even.
So I’m comfortable with the theme and I’m careful with the entry. That’s the whole game.
How to trade it
Here’s the explainer, the way I’d think about it with a trader I’m mentoring:
Pick your instrument. Nasdaq futures are the cleanest proxy for the AI theme. ES gives you the wider market. I’m not saying trade the NQ but at times it is a proxy.
Know the catalysts. Megacap earnings, Nvidia’s results, changes in capex guidance, interest rate decisions and big macro data. A raised capex guide and a cut capex guide tell very different stories. Big IPO pricing dates are now catalysts too.
Trade the structure, not the story. Narratives are easy and prices are hard. I look at where value is accepted and rejected, using Market Profile and key levels, instead of arguing about whether AI is real.
Respect event risk. Size down into big prints and listings. Gaps and volatility can hurt.
Plan both directions. Know what you’ll do if it keeps running and what you’ll do if it corrects. Don’t marry a theme.
Keep stats. If you can’t say which setups make you money in this market, you’re guessing.
The common mistakes I see are confusing a long-term theme with a short-term entry, averaging down on a falling position, and sizing up because the story feels certain.
The bottom line
The AI trade isn’t the dot-com bubble, but it isn’t risk-free either. Strong earnings and lower multiples are a better foundation than 2000 had, yet concentration and spending are heading in an uncomfortable direction, and the IPO wave is a reminder of how much enthusiasm is already priced in.
If you’d like help building a process for trading this market, I’m offering a free 15-minute call. Email [EMAIL] and I’ll send the booking link.
Dan Goldberg, Goldberg Trader Development & Mentorship
Educational content only. Not financial advice. Trading leveraged products carries a high level of risk and may not be suitable for everyone.
