Don’t let the AI “bubble” put you off using AI

Oracle signs a $300bn deal with OpenAI. A vast AI data centre project in Texas called Stargate promises gigawatts of computing power. OpenAI commits hundreds of billions of dollars to cloud providers. Commentators talk about trillions of dollars of AI infrastructure spending.

And then? Oracle’s share price falls. Other stocks wobble after announcing AI partnerships. Suddenly the hot story is whether there’s an AI bubble.

If you’re an ordinary professional just trying to do your job better, this can sound like a warning: “If Wall Street thinks AI is risky, maybe I should stay away too.”

That’s exactly the conclusion you shouldn’t draw.

There is a bubble forming – but it’s mainly in how much money is being thrown at data centres and infrastructure, not in the everyday usefulness of AI for people like you.

Two very different AIs

Let’s separate two things that get muddled together:

1. Investment & infrastructure AI
   This is the world of $300bn cloud deals, multi-gigawatt data centres and highly leveraged balance sheets. It’s where Oracle, OpenAI, Amazon, Microsoft and others are racing to lock in future computing power.
2. Everyday, practical AI
   This is you using AI to draft emails, build a report, clean up a spreadsheet, summarise a meeting, or turn a messy idea into a clear plan.

The same underlying technology sits under both – but the risk profile is completely different.

• Investment AI is about, “Will this trillion-dollar capex race pay off?”
• Everyday AI is about, “Can I get an hour back this afternoon?”

You don’t need to bet your mortgage (or your company’s balance sheet) to get the second one.

What’s actually going on with the “bubble”

A quick decode of the big headlines:

• OpenAI has committed to spend hundreds of billions of dollars on cloud computing over the next few years, including a $300bn deal with Oracle starting later this decade.
• Projects like Stargate in Texas and other new data centre sites are designed to power future, more powerful AI models with enormous energy and chip requirements.
• To make this happen, companies are taking on huge capital expenditure and, in some cases, large amounts of debt – years before the full revenue shows up.

From an investor’s point of view, it’s a fair question:

“Are these infrastructure bets sensibly priced, or are we in a classic tech overbuild?”

That’s where talk of an AI bubble comes from:

• Share prices that jump whenever an “AI deal” is announced – and then start to fall back once everyone realises the scale of the bill.
• Companies becoming heavily dependent on a small number of AI clients for future growth.
• Massive spending now, with the payoff uncertain and years away.

This is all real. If you’re buying the stocks or lending the money, you should care deeply about it.

But none of that automatically says, “AI isn’t useful.” It says, “Some of the ways we’re financing and pricing AI infrastructure might be over-excited.”

The bit that’s not a bubble: your time

Now flip perspective.

Imagine you’re not Oracle, OpenAI or a hedge fund. You’re a manager, analyst, founder, consultant, accountant, HR lead, or ops person.

What does AI look like for you?

It looks like:

• Turning a rough bullet list into a clear first draft of a document in minutes.
• Summarising a 20-page PDF or a 60‑minute meeting into the 10 lines you actually need.
• Generating five options for that awkward client email so you’re not staring at a blank screen.
• Cleaning and restructuring spreadsheet data that would normally take you an afternoon.
• Turning messy notes into a project plan with owners and timelines.

For you, the important questions are:

• Does it save me time?
• Does it improve the quality of my output?
• Is it safe and sensible to use for this task?

If the answer to the first two is “yes”, and you’re not breaking any data or confidentiality rules, then the fact that some giant data centre might be overfunded is… largely irrelevant.

You’re renting intelligence by the minute. You are not underwriting a $300bn data farm.

Why this matters right now

We’re at an odd moment:

• On the financial side, we may well be entering the “hangover” phase of the first AI investment mania. The easy trade – announce an OpenAI-style deal and watch your stock jump – is fading.
• On the practical side, AI has quietly matured to the point where it can genuinely remove hours of low‑value work every week for a typical professional.

Those two things can be true at the same time.

Historically, that’s often what happens with transformative technologies:

• The investment cycle swings wildly – bubbles, corrections, restructurings.
• The everyday value keeps compounding – quietly, in the background, on Tuesday afternoons.

Railways, telecoms, the early internet: in each case, investors overbuilt at some point. Some big players got burned. But the people and organisations who learned to use the infrastructure carried on winning for decades.

AI is lining up the same way.

So what should an “average Joe” actually do?

If you’re feeling put off by the AI bubble talk, here’s a simple reframing:

Let the big players worry about whether they’re overpaying for data centres. Your job is to decide whether AI can give you time and headspace back.

In practical terms:

1. Pick one painful task
   Choose something routine and low‑risk: drafting emails, writing meeting notes, summarising long documents, planning your week.
2. Use AI as a first draft, not a final answer
   Let the model get you from 0 to 70%. You stay firmly in charge of the last 30%.
3. Measure the time saved
   Don’t rely on vibes. If a task used to take 60 minutes and now takes 20, that’s real value – whatever the market cap of Oracle happens to be today.
4. Build a small personal toolkit
   A writing assistant, a data helper, a meeting summariser and a research companion already cover a huge chunk of daily knowledge work.
5. Stay curious, not cynical
   Healthy scepticism is useful. Cynicism – “it’s all a bubble so I’ll ignore it” – is how people miss the upside of major shifts.

The bottom line

When you see headlines about $300bn deals and trillion‑dollar AI capex plans, remember they’re describing the plumbing, not the value you personally can unlock.

Yes, there’s a real debate to be had about whether AI infrastructure spending is overheating. Yes, some investors, lenders and even big tech firms will get their timing wrong.

But that shouldn’t stop you from:

• Getting your evenings back from pointless admin.
• Turning around high‑quality work faster.
• Using AI as a practical tool, not a philosophical argument.

The investment bubble will inflate and deflate. The question for you is much simpler:

“Am I using the tools that are already here to make my working life better?”

If the honest answer is still “not really”, the headlines aren’t your biggest problem.