Just How Big Is AI? I Tried Explaining It to My Wife

Donald Trump has a new name for it – SI.

Whatever you call it, AI has dominated more headlines than any other topic in the world for more than a year now.

Despite hearing about it so often, I feel I know so little about it.

How big is it REALLY?

What are its different parts? How do they fit together?

And why do I keep hearing people say that India has no real AI play?

So, I did some research and put together a few pointers with two things in mind. They should be easy to understand and generally accurate.

1. How big is AI?

The AI market cap is currently around $25tn.

For perspective, India’s annual GDP is roughly $4 trillion. China’s is around $20 trillion.

While this is not an apples-to-apples comparison, it can be said that the only economy larger than the AI economy right now is the USA.

That is the crazy scale.

And when we realise that all this has happened in last 2-3 years.

That is the crazy rate of growth.

So what exactly is happening in something that is so massive?

2. Frontend of the Miracle = AI Labs & LLMs

At the front end are AI labs such as OpenAI and Anthropic.

These make LLMs like ChatGPT and Claude. (LLM = Large Language Model)

These companies develop the models that power AI tools or magic tool that is leading this miracle.

But these AI labs or LLMs need other components to work.

3. Parts = GPUs & Memory

Think of it this way. A program needs a computer to run. A computer has processors and memory to run on.

AI models work on more or less same basic principle, but at a vastly larger scale. Remember $25tn?

This is where GPUs and memory come in.

GPUs are specialised processors that handle the massive calculations required to run AI models.

NVIDIA is the dominant name in GPUs.

And it went from being the 800th most valuable company in 2015 to the #1 most valuable company in 2025.

But NVIDIA does not manufacture all its chips itself.

It relies on manufacturers such as Taiwan’s TSMC and memory makers such as South Korea’s SK Hynix.

How big is Hynix? Hynix is now valued at 25% of the ENTIRE SOUTH KOREAN stock market.

Crazy scale. Crazy rate of growth.

4. Aggregators of These Parts (or Hyperscalers)

So we have the processors, RAM & the program. But no computer yet.

Here come the companies that buy vast quantities of computing hardware from NVIDIA, connect then using sophisticated code and rent this enormous computing power out to AI labs.

These are the existing cloud companies: Microsoft, Amazon and Google.

These are now called hyperscalers (reminding again of crazy rate of growth)

Why? Because much of the money being pumped into AI infrastructure comes from these companies.

Why? Because they are some of the most cash-rich companies in the world.

Why are other cash-rich companies, say oil companies, do this? Because it requires sophisticated expertise to prepare these supercomputers which they learnt in their cloud business.

Hence, the hyperscalers are usually cloud giants.

And it is these hyperscalers that are leading the crazy rate of growth.

5. Where Is All of This Magic Happening?

So, we have the program, the computer, the processors, memory, everything. Whats left?

Office. Or Home. Whatever you like to call.

These mega computing systems need a place to operate. Power to run on. That home is a data centre.

A data centre is a facility that houses all these computing hardware, networking equipment and other infrastructure.

This where LLMs are training 24×7.

On the data fed by peoples usage.

This is where this scary monster becomes more scary.

6. More On Data Centers

Data centers need three things in principle and in enormous quantities:

• Land to build the facility.
• Reliable electricity to run the machines.
• Cooling systems to prevent the equipment from overheating.

This creates another business opportunity.

And while it may sound like just another cog, remember that the monster machine is worth $25tn. Every tiny cog is worth billions here.

So, how big are data centres?

$500bn. For reference, the global mobile phone market, including Apple, Android and others, is worth $600bn.

Notice that India has no visible presence in any of the layers mentioned above. This is where it wants to get its foot in the door.

In Part 2, we will look at India’s data centre opportunity, the infrastructure challenges it faces and the environmental cost of this boom.