Is AI in a bubble (Part 1)?

The bottleneck isn't what you want to say — it's how long it takes to type it. Wispr Flow removes the bottleneck.

Speak naturally and get polished, send-ready text for executive summaries, client updates, board recaps, investor notes, or just the 30 Slack messages you're behind on. Flow strips filler, formats numbers and lists, and preserves your tone.

Used by teams at OpenAI, Vercel, and Clay. 89% of messages sent with zero edits. Works in every app on Mac, Windows, and iPhone.

Start flowing free

There are moments in investing when a single technological breakthrough captures the world's imagination.

Railroads transformed the nineteenth century. The internet reshaped the late 1990s. Smartphones revolutionized the past fifteen years.

Today, that technology is artificial intelligence.

Almost every conversation in financial markets eventually leads back to AI. Investors are pouring capital into anything remotely connected to it. Companies are reorganizing their businesses around it. Governments are racing to build the infrastructure needed to support it. Every earnings call, investment conference, and analyst report seems to arrive at the same destination.

The question is no longer whether AI will change the world.

The real question is whether investors are paying a sensible price for that future.

History teaches an important lesson that markets often forget: transformational technologies and investment bubbles are not mutually exclusive.

The internet changed the world.

Railroads changed the world.

Electricity changed the world.

Yet each of these revolutions experienced periods when investment flowed into the sector much faster than the economics could justify.

So, are we witnessing another technological revolution that will create enormous long-term wealth, or the formation of a classic capital-cycle bubble?

The answer isn't straightforward.

Before we can assess whether AI is becoming a bubble, we first need to understand what companies are actually spending hundreds of billions of dollars on.

On June 25, 2026, Apple made a highly unusual move.

Without launching a major new product or hardware refresh, the company raised prices across several product lines.

Apple attributed these increases to soaring memory chip prices, noting that it had never experienced such dramatic component cost inflation in such a short period.

That announcement revealed something many investors had overlooked.

The AI boom wasn't simply inflating stock prices.

It was beginning to create inflation in the real economy.

Consumers weren't paying more because Apple wanted larger profit margins—they were paying more because advanced memory chips had become scarce as AI companies purchased virtually every available supply.

The easiest way to understand the scale of today's AI race is to ignore stock prices entirely and focus instead on capital expenditure.

Before ChatGPT arrived, the combined annual capital spending of Amazon, Microsoft, Google, and Meta totaled roughly $90 billion.

Then everything changed.

Most AI resources are scattered, untested, or buried behind a paywall.

The Shift puts it all in one free newsletter, covering what’s new, and what practically works under 5 minutes a day.

A vault of 3,000+ vetted tools so you’re never guessing. A 1000+ prompt library for every use case you’ll actually hit and free AI courses that make sure nothing stays theoretical.

Plus, right now, 3 subscribers win a free 1-year Claude Pro subscription just for signing up.

Subscribe to The Shift

In just six years, annual investment increased roughly eightfold.

That isn't normal.

Companies rarely accelerate infrastructure spending at this pace unless they believe they are standing in front of a once-in-a-generation opportunity.

Nearly every additional dollar is flowing toward one objective:

Building AI infrastructure.

When people think about AI investment, they often picture software.

Large language models.

Chatbots.

Coding assistants.

Image generators.

But those applications are only the visible layer.

Beneath them sits an enormous industrial infrastructure that looks far more like a utility company than a software business.

Every time you ask ChatGPT a question, your phone performs very little of the actual computation.

Instead, the work happens inside massive data centres filled with thousands of specialized processors.

A modern AI data centre can contain around 100,000 Nvidia GPUs.

Each GPU costs approximately $30,000–$40,000.

That means the processors alone can represent an investment of $3–4 billion.

Then come the additional costs:

By the time everything is included, a single AI data centre can cost anywhere between $10 billion and $25 billion.

Suddenly, the scale of today's investment becomes much easier to understand.

These companies aren't buying software licences.

They're building industrial infrastructure comparable to power stations.

The answer comes down to data.

Global data creation has exploded over the past decade.

Around 2010, the world generated roughly 2 zettabytes of data annually.

By 2026, that figure is expected to reach 221 zettabytes.

Whether every projection proves accurate or not, the overall trend is undeniable.

Every search.

Every video.

Every email.

Every software application.

Every AI interaction.

All of it requires computing power.

This explosion in digital information is the primary justification behind today's infrastructure spending.

The investment thesis is simple.

If computing demand continues growing at this pace, today's expensive infrastructure could eventually look remarkably inexpensive in hindsight.

That is the bet Big Tech is making.

One of the most striking figures comes from research by PIMCO.

Over the next two years, the largest technology companies are expected to spend approximately 94% of their operating cash flow on AI-related capital expenditure.

Think about what that means.

Imagine a business earning $100.

Traditionally, management would allocate those profits across research, employee compensation, dividends, share buybacks, acquisitions, and new products.

Instead, nearly $94 out of every $100 is being reinvested into AI infrastructure.

Only six dollars remain for everything else.

Just a few years ago, that figure was closer to 40%.

The shift is extraordinary.

Very few industries have ever concentrated capital this aggressively around a single technological assumption.

Everything ultimately rests on one belief.

The world's largest technology companies believe future demand for AI computing will be so immense that today's spending will eventually appear conservative.

If they're right, today's investments could generate exceptional returns for decades.

If demand grows more slowly than expected, however, enormous amounts of capital could remain underutilized.

And that is where the debate becomes much more interesting.

Building infrastructure is one challenge.

Generating sufficient economic returns from that infrastructure is another entirely.

Imagine spending $10 million to build a factory that manufactures coffee machines.

After a full year, the factory generates just $400,000 in annual revenue.

Would anyone build a second factory?

Probably not.

The issue isn't whether people drink coffee.

The issue is whether demand justifies the investment.

AI now faces the same economic question.

Few people doubt that AI has transformative applications.

Few doubt that the technology will continue improving.

The debate is about economics.

Can AI generate enough revenue to justify the unprecedented infrastructure currently being built?

Investors often confuse technological success with investment success.

History repeatedly shows they are not the same thing.

Claude vs Gemini. GPT-7 vs Llama 5. Which AI lab ships AGI first. These are live Kalshi markets with real money on both sides, updated in real time as releases land. The person who follows model cards and tracks evals has a genuine edge here. If that's you, trade it.

Trade the model race

Looking at the numbers from a purely investment perspective, analysts have asked a simple question.

If investors expect roughly a 10% annual return on the hundreds of billions being invested, how much annual revenue must AI generate to justify today's spending?

The answer is approximately $650 billion per year.

Now compare that with current estimates.

Compare those figures:

The gap is impossible to ignore.

For every dollar AI currently generates, technology companies appear willing to invest nearly ten dollars building future capacity.

This widening disconnect has become one of the defining questions facing investors today.

Not because AI lacks promise.

But because nobody yet knows who will ultimately generate the profits required to justify today's extraordinary spending.

In Part 2, we'll move beyond infrastructure and explore what is actually happening inside businesses adopting AI—where early results are proving far more complex than the market's enthusiasm might suggest.

Read more of our articles

Disclaimer: The information provided on this website is for educational and informational purposes only and does not constitute financial, investment, or trading advice. Investing in securities involves risk, including the potential loss of principal; always conduct your own research and consult a qualified financial professional before making investment decisions.