Every Top AI Company Is Racing to Be Worth a Trillion Dollars. History Says Most Will Lose.

The last time money rushed into a technology like this, the winners were not the companies everyone was funding.

Start with the ghost story

In the late 1990s, the hottest company on earth was one most people today have never heard of. Nortel built the physical backbone of the internet, the cables and switches that carried three out of every four bits of North America's web traffic. Its revenue once jumped 133 percent in a single year. At its peak it was worth more than $360 billion, making it one of the most valuable companies in the world. In 2009, it filed for bankruptcy. The technology it built did not fail. It succeeded so completely that it became cheap, ordinary, and replaceable, and the real fortunes went to the companies that built on top of it: Google, Amazon, Netflix. Hold that story in your head, because it may be happening again, right now, with AI.

The race defying all normal logic

Today the Nortels of our moment are OpenAI and Anthropic, the makers of ChatGPT and Claude. They are sprinting toward valuations near a trillion dollars each, and the speed is hard to believe.

• Anthropic's valuation went from $380 billion to $965 billion in about three months. That final number is bigger than the entire yearly economy of Saudi Arabia.
• Anthropic and OpenAI together swallowed roughly 14 percent of every venture-capital dollar raised on the planet last year.
• Anthropic's sales went from about $9 billion to $47 billion a year in roughly five months. The software giant **Salesforce took two decades **to travel that far.

These are companies that sold almost nothing two years ago and still lose money today, valued like the most important businesses on earth.

Why they are all sprinting at once

The reason is a single shared belief: that AI will become as basic as electricity, and that only a few companies will own the most powerful version of it. If that is true, the prize for winning is almost unlimited and the price of falling behind is becoming irrelevant. So the labs are trapped in a race where slowing down feels like dying. Each raises a giant pile of money, spends it instantly on computing power and talent, and has to raise again to keep up with the others. The valuations are not really prices. They are bets on who will still be standing when the dust settles.

The demand is real, and that is the trap

It would be easy to call this pure hype, but the demand underneath is genuine, which is exactly what makes it seductive. Businesses are not toying with these tools. They are wiring them into daily work, and Anthropic has even had to ration how much customers can use Claude at peak hours because it cannot supply enough. Real companies, paying real money, asking for more than the supplier can give. That is what convinces investors to keep doubling down. And it is the same thing that made Nortel look invincible right before the fall, because soaring demand for the plumbing is not the same as owning what flows through it.

Soaring demand for the plumbing is not the same as owning what flows through it.

Why the winners may not be the favourites

Here is the twist that should make you sit back. The cost of using these AI models has collapsed, falling hundreds of times over in just a few years. When the core technology gets that cheap that fast, it stops being the thing worth the most money. The advantage slides to whoever builds the most useful product on top of it. That is precisely what happened with the internet: the fibre got cheap, and the value floated up to Google and Amazon. If the pattern holds, today's trillion-dollar labs are this era's backbone builders, essential and dazzling and quietly destined to be overtaken by companies that are still tiny, or do not exist yet.

The one line to carry

So when someone tells you OpenAI or Anthropic has won the AI race, remember the ghost of Nortel. The race is real, the growth is real, and the outcome is genuinely open. The thing to watch is not the next record valuation, because those will keep coming. It is whether these companies can build something customers cannot walk away from the moment a cheaper version appears.

The companies that define the AI age may not be the ones devouring 14 percent of the world's investment today. They may be the ones quietly building on top of it, the way Google once built an empire on fibre that someone else went bankrupt laying.