The Possible Collapse of AI
- Ben Clarke
- Oct 1
- 3 min read
Updated: Oct 4

Not personalized financial advice - your money, your choice
Circles are irrational - an irrational number literally defines them. Nvidia’s $100B bet on OpenAI? Even more irrational.
It’s like giving your cousin 20 bucks for lunch, only for him to use it to buy your old PlayStation immediately. Sure, you made a sale—but did anything actually change? Did he even get lunch?
Why do I find this completely irrational? As part of the agreement, OpenAI has agreed to purchase $350B of compute. Sounds a little circular, no?
If I could invest in a company only to have them come back to me and spend the money I just gave them with my company, I would do that deal as many times as possible. And that’s exactly what companies did during the dot-com bubble.
Lucent was like Nvidia—supplying the hardware everyone needed. To juice sales, they started lending billions to smaller telecom players such as Winstar. Those companies then spent the borrowed money… on Lucent’s equipment. On paper, Lucent looked like it was booming. In reality, it was just selling to customers who couldn’t afford to pay by giving them its own money to pay with.
When Winstar went bankrupt, Lucent was left holding the bag. Its stock collapsed from $84 in 1999 to just 56 cents by 2020. Tens of thousands of employees were laid off, and Lucent spent years selling off pieces of itself to stay alive.
Nortel pulled the same trick. In 1998, it bragged about a $170M financing deal with Net2000—money Nortel had essentially loaned them, just so Net2000 could turn around and buy Nortel’s gear. It made the market look strong, but the foundation was rotten.
When the dot-com bubble burst, Nortel’s customers couldn’t pay their debts. The company unraveled, eventually sliding into bankruptcy.
History doesn’t always repeat, but it does rhyme. Lucent and Nortel rapped this rap more than 20 years ago. Nvidia’s not doomed—but the rhymes are rhyming.
Nortel died a slow and painful death over a few years, and Lucent managed to have a slightly better outcome. The stock went’ t down from $84 in 1999 to 56 cents by 2020. There were then many tens of thousands of employees who were laid off, and they scrambled to raise cash by selling off business units.
In 2003, they merged with similarly distressed Alcatel, but in 2024, the SEC fined Lucent $25M for its lack of cooperation in a fraud case. B
By 2006, Alcatel gobbled them up, but the corporate cultures of the companies didn't mesh well. In 2016, it was sold to Nokia, where Nokia then fully absorbed the company and ended the Lucent brand.
Again, this doesn’t mean that we’re at the top of the market; necessarily, it took 3 years of gradual decline before these companies hit rock bottom. However, it does make me want to watch out for more of these ‘circular’ deals.
Prof G Markets flagged this as an issue and created the visual for this post, so huge credit to them there, and check out their pod on this topic here. I could not recommend their podcast more!
If you learned something, subscribe to my YouTube channel or newsletter. Think of it as a healthier kind of circular deal—I write, you read, we both come out ahead.”







