Why does everyone keep planning for 2030?

Why does everyone keep planning for 2030?

Look around.

Governments are publishing AI scenarios built around 2030.

Google says it wants its operations running on 24/7 carbon-free energy by 2030.

The UK has committed up to £2 billion to public compute infrastructure through 2030.

AI infrastructure spending estimates now stretch into 2030 and beyond.

And the companies building the hardware are locking themselves into hundreds of billions of dollars of supply and capacity commitments years into the future.

That doesn’t prove somebody has a secret 2030 master plan.

But it does raise a better question:

Why did 2030 become the year everybody is building toward?

Because once enough powerful institutions choose the same deadline, the deadline starts changing behavior.

A government sees 2030.

A utility sees 2030.

A data-center developer sees 2030.

A chip company sees 2030.

A bank financing the project sees 2030.

An investor valuing the company sees 2030.

Then they all start spending today based on what they expect the world to look like four years from now.

That’s how a forecast can become an engine.

Look at the money involved.

One recent analysis estimates U.S. AI infrastructure investment rising from roughly $581 billion in 2026 to more than $1 trillion annually by 2030.

The Financial Times reports that the world’s largest technology companies have already raised roughly $500 billion this year to finance AI infrastructure, with the borrowing expected to grow substantially.

And the physical footprint is no longer theoretical.

There are nearly 100 U.S. hyperscale data centers operating, about 120 under construction and roughly 460 more planned, according to a Washington Post analysis of satellite imagery. Some consume at least 100 megawatts.

Now comes the part that makes the whole thing stranger.

The people spending this money don’t actually know what 2030 looks like.

The UK government’s own AI scenarios explicitly say they are not predictions. One scenario has AI automating most remote human work from 2029 onward. Another assumes much slower development.

So we have a peculiar situation.

Nobody can tell us exactly what AI will look like in 2030.

Yet hundreds of billions of dollars are being committed against that date.

That is where I get suspicious.

Not suspicious in the “secret room with 12 billionaires” sense.

Suspicious in the much more boring financial sense.

Who decided the future had a deadline?

Because once everybody builds for the same future, somebody eventually has to be right.

If AI demand explodes, today’s infrastructure spending may look cheap.

If demand disappoints, the infrastructure doesn’t disappear.

The debt remains.

The power plants remain.

The transmission lines remain.

The buildings remain.

The chips become obsolete.

And the contracts remain.

This is why the current debt market deserves attention.

Reuters reported that investors are already becoming more selective toward AI-related corporate debt. AI bond spreads were running around 115 basis points versus 78 basis points for investment-grade debt overall, while AI-related issuance is expected to reach roughly $420 billion next year.

The financing machine is therefore expanding at the same time the people providing the financing are asking harder questions.

That creates the part nobody can answer yet.

What happens if 2030 arrives and the world built for doesn’t?

Because the most expensive mistake in history would not be predicting the future incorrectly.

It would be building the future first and discovering afterward that nobody needed quite as much of it as expected.

And there is one more thing worth watching.

The physical world is already forcing the issue.

Data-center demand is growing faster than some power systems can accommodate. Companies are increasingly building their own generation, negotiating long-term power contracts and moving projects toward locations where electricity can actually be delivered. The Washington Post’s satellite analysis shows how rapidly these facilities are spreading into rural areas, while research published in Nature estimates AI data centers could consume around 1% of global electricity by 2030.

So maybe 2030 isn’t a prophecy.

Maybe it’s something more powerful.

A deadline that enough people believe in that they are collectively spending hundreds of billions of dollars trying to make it real.

That’s how a future stops being a forecast.

It becomes an investment position.

And if everyone is positioned for the same future, the uncomfortable question isn’t what happens if they’re right.

It’s what happens if they’re wrong.