This post was authored by Adem Tumerkan, Dunham's Content Editor. If you have questions concerning today's topic, please call us at (858) 964 - 0500. Hold us to higher standards.

Author Update – November 8, 2025:

Since publishing three weeks ago, the mountain of corporate debt tied to the AI has surged and began raising alarms. Via The New York Times recent report, borrowing is now “crowding out” cash-funded growth.

Key Takeaways:

• AI Mania: The excitement around artificial intelligence is starting to look a lot like past market bubbles.
• Real Returns: Despite the hype, most companies aren’t seeing the profits they expected from AI adoption.
• Big Spending: Tech giants are pouring trillions into AI, but more of it now comes increasingly from borrowed money (be wary of debt-fueled growth).
• Supply Crunch: Shortages in power, chips, and materials today could turn into costly gluts tomorrow due to overcapacity and excess competition.
• Market Reality: AI won’t collapse because it failed — it’ll cool off because success went too far, too fast (as most booms suffer from).

The AI Bubble Is Swelling: Debt-Fueled Growth and Hype Are Flashing Warning Signs

Since ChatGPT went public in late 2022, artificial intelligence (AI) has dominated every headline, every conversation, and every investment theme.

We’ve heard it all - “AI will take over the world,” “AI will reshape the economy,” and “Anyone not investing in AI is a fool."

In fact, NVIDIA recently hit a $5 trillion market cap - making it (in dollar terms) larger than Germany, Japan, or India¹.

And while there’s significant merit for the upside of AI, I’m not convinced it’s the end-all, be-all the crowd expects.

If history means anything, the success and hype surrounding AI are already planting the seeds of its own demise.

So the real question becomes: Is this another hype cycle - like the roaring 1920s, the dot-com bubble, or the housing boom?

Many may believe that this time is different. 

But, as Sir John Templeton warned, “The four most expensive words in investing are: This time is different.”

Three Forces Fueling the AI Boom

I believe three things are carrying AI right now:

• Hype – investors excessively cheer on every new AI deal, earnings beat, and optimistic forecast.
  
  
• Corporate spending - Companies are pouring record sums into chips, servers, and data centers, increasingly turning to debt to keep the momentum going.
  
  
• Infrastructure shortages - Giving power producers, commodity suppliers, and construction firms the kind of pricing power they haven’t seen in years

Of course, at first glance, these sound like tailwinds. And they certainly have been.

But remember. In markets, what fuels the boom cycle often ends it.

• Hype inflates expectations faster than reality can deliver.
  
  
• Corporate spending eventually becomes debt-fueled and increases default risk.
  
  
• And shortages - whether they’re chips, power, or capacity - eventually invite oversupply.

Those are the same ingredients that fueled the dot-com bubble, the housing boom, and every mania before it.

• Read my piece “How Debt Fueled History’s Greatest Bubbles and Market Crashes” to learn about some of history’s lesser-known booms (and the busts that followed).

So, let’s take a closer look at what’s really behind the boom - and why it may not last.

The AI Market Hype: Overbought or Just Right?

We all know that emotions drive markets.

• Greed lifts prices beyond reason
• Fear brings them crashing back.

And right now, markets are drunk on optimism about AI.

Analysts, CEOs, investors, entrepreneurs – all convinced that AI will change everything and mint fortunes along the way.

And so far, they seem right.

But what if it doesn’t hold? What if the promised productivity revolution never quite materializes?

Well, if that happens, trillions in paper wealth could vanish – fast – as hype turns to fear.

And that risk is growing.

Back in August, the Massachusetts Institute of Technology (MIT) published a worrying paper showing that most companies aren’t seeing the revenue boom AI was supposed to deliver.

According to The GenAI Divide: State of AI in Business 2025², only 5% of AI pilot programs achieved meaningful revenue growth. The other 95% stalled, delivering little measurable impact on profit. The researchers found the problem isn’t exactly the models – but rather the execution (aka companies are overspending on tools that don’t integrate, don’t learn from workflows, and don’t scale).

The result is an economy where AI adoption is soaring, but returns aren’t.

Even worse is that spending could start to backfire. If companies keep pouring money into underperforming systems, the payoff gap will grow wider - and valuations will have to adjust (more in the next section).

Still, falling hype doesn’t mean AI is useless (far from) - just that it might be overpriced right now.

For example, as Gartner’s Five Phases of the Hype Cycle shows, technology booms often follow the same pattern:

1. Technology Trigger: A breakthrough sparks excitement - long on promise, short on proof.
   
   
2. Peak of Inflated Expectations: Hype completely outruns fundamentals.
   
   
3. Trough of Disillusionment: Projects start to fail, optimism fades, and only the strongest survive the culling.
   
   
4. Slope of Enlightenment: The survivors consolidate, learn, refine, and rebuild.
   
   
5. Plateau of Productivity: Real value emerges, albeit at a much slower and saner pace.

Figure 1: Gartner, Dunham, 2025

I’d estimate we’re around the peak of inflated expectations and sooner than later will plunge into the trough of disillusionment.

Hype seems surreal in this market. But that can quickly unwind.

Corporate Spending: The Rise of Debt-Fueled AI Growth

Big tech is spending fortunes in this AI arms race.

To put this into perspective, Alphabet, Microsoft, Amazon, and Meta are projected to spend nearly ~$400 billion this year on AI-related items³.

• That’s more than the European Union’s entire annual defense budget.

And the spree isn’t even close to slowing. . .

Morgan Stanley estimates nearly $3 trillion in AI infrastructure spending from 2025–2028, enough to add about 0.5% to U.S. GDP growth this year and next.

Figure 2: UnderstandingAI, Substack, October 2025

Such massive investments are helping push companies’ profits (and their share prices) to records.

But this will – inevitably – lead to a significant squeeze in margins.

Why? Because of the law of diminishing returns.

• Meaning every boom begins with productive spending - and ends with unproductive spending.

The early money spent in AI created breakthroughs - ChatGPT, copilots, image generation, and autonomous code. Each dollar then delivered exponential returns in capability.

But now? The spending curve has gone parabolic while the innovation curve is flattening – aka more spending for less innovation.

Said another way - the marginal gain from each new model or chip is shrinking - yet the cost of chasing it keeps drastically rising.

• It’s like squeezing juice from an orange - the first press gives you everything, the next gives you less, and eventually, you’re just twisting the rind and hurting your hand for nothing.

Figure 3: Dunham, 2025

Simply put, the weight of this enormous AI spending will eventually collapse under its own mass.

Meanwhile, the top players in the AI space are increasingly engaging in sketchy circular financing deals - which is historically a fragile sign.

• What’s circular financing? I wrote about this before, but in short - it’s when suppliers essentially fund their own sales. GPU makers, for example, are reportedly lending money to data-center firms so those firms can buy more chips - booking revenue up front while financing future buying. It’s a loop that looks like growth but risks turning into malinvestment - much like the late-1990s telecom boom when Cisco and Lucent extended credit to customers to sustain orders. And when the lending stopped, so did the buying.

The same risk now looms over AI.

Because subsidized momentum fueled by self-financed spending = more fragility.

• Goldman Sachs reports that AI-infrastructure companies have issued about $140 billion in new corporate debt in 2025, already surpassing last year’s total⁴.
  
  
• JPMorgan Chase estimates that debt tied to AI has ballooned to $1.2 trillion, making it the largest segment of the investment-grade market^5.
  
  
• A massive $30 billion data center financed through private debt and largely outside traditional credit markets marks a new phase of AI growth - fueled less by cash and more by “shadow” financing⁶.

What started as a cash-flow story is now entering its credit-driven chapter.

Keep in mind that most companies here remain financially strong, but the trend shows that AI spending is stretching even Big Tech’s balance sheets.

Thus, as credit replaces cash, the story of organic, self-funded growth is giving way to one of rising leverage, opaque financing loops, and uncertain payback - the classic signals of a boom nearing its peak.

Put simply, every boom runs on hype until it starts borrowing against it. Then it’s a recipe for disaster.

Supply Crunches Today, Overcapacity Tomorrow

A ripple effect from all this AI hype and spending has been the extreme shortage of supplies - from commodities, power, and space.

• The average hyperscale data center now consumes around 500,000 to 1 million gigajoules of energy per year⁷ - roughly the same as 50,000 U.S. homes.
  
  
• Building one requires tens of thousands of tons of steel, concrete, and copper - enough copper alone to stretch more than 1,000 miles of wiring⁸.
  
  
• Water usage for cooling and humidity control has surged from 6 billion gallons in 2014 to 17.4 billion in 2023⁹.

Because of this, prices for copper, aluminum, and energy have soared, and some regions are already rationing electricity for new projects.

Many Americans already dislike the idea of AI taking their jobs. And they’ll like it even less when it starts raising their power bills further.

Figure 4: WIRED.com, October 2025

Thus, what started as an AI revolution is now colliding headfirst with the limits of the physical world and rippling into something far bigger.

And when prices rise, entrepreneurs rush in to chase profits.

But – like every bubble before - we’re not seeing the true innovators anymore.

We’re seeing what I call “slop-preneurs” - the latecomers who pile in when the real opportunity’s already been picked clean.

• They put AI in their name and expect markets to love it.
• They build when materials are most expensive.
• They borrow when credit is tightest.
• And arrive just in time to catch the downslope.

Thus, they’re not creating efficiency - just chasing momentum.

And in every cycle, it’s the slop-preneurs who show up last - and leave first.

But - there’s upside when they do show up. . .

Because these slop-preneurs will bring on extra supply – helping ease shortages and push prices down.

Sure, many will likely go bust from excessive competition and falling prices that crush margins (from overcapacity). But the last firms standing will pick over their bones and grow stronger.

The only problem is that this may take a while to play out.

• You can’t build a nuclear reactor, data center, or new copper mine overnight, right?

The point is, shortages always create gluts, and gluts always crush margins.

This boom has already planted seeds of its own destruction.

• I wrote more about this dynamic in my February piece, “The Capital Cycle: Do Supply Trends Drive Markets More Than Demand?” - showing how this boom-bust pattern plays out over time.

Final Thought: The AI Bubble Is Poised To Pop

Am I saying AI isn’t important? No - it’s clearly revolutionary.

Am I saying it’s not worth watching? Absolutely not. It will reshape industries, workflows, and maybe even economies.

But is the hype worth the price being paid for it right now? That’s where I have doubts.

• It’s like climbing a mountain long after the summit’s been reached - the view doesn’t get better, the air just gets thinner.

That’s where we are with AI. The technology isn’t the problem. The hype and valuations are.

History shows that cycles – booms and busts – are inevitable. And every great innovation eventually overshoots its value before crashing back down into reality.

• The internet did.
• Railroads did.
• Electricity did.

And I believe AI will too.

And when it does, it won’t mean the end of AI - far from it. It’ll simply mark the end of the illusion that it could defy gravity at any cost.

Thus, AI won’t collapse from failure - it’ll stumble under the weight of its own success.

Because in markets, progress is never the real problem. Hype, valuations, and scarcity always are.

But as always, this is just some food for thought.

FAQ:

Is the current AI boom turning into a bubble?
Many analysts believe the 2025 AI boom shows early signs of a market bubble. Rapid valuations, hype-driven investment, and heavy corporate borrowing mirror past cycles like the dot-com era. While not collapsing yet, AI markets may be approaching a cooling phase.

What do diminishing returns mean for AI investments?
Diminishing returns in AI mean each new dollar invested creates less innovation or profit. Early breakthroughs like ChatGPT fueled exponential gains, but now companies are spending billions for smaller improvements. This signals that AI growth may be shifting from explosive innovation to efficiency and consolidation.

How can investors prepare for a potential AI market slowdown?
Diversify across industries using AI, not just building it. Prioritize companies with strong cash flow, healthy balance sheets, and measurable returns from AI integration. Avoid firms relying on excessive debt or hype to sustain growth.

Will AI innovation stop if the bubble bursts?
No — innovation won’t end if the AI market corrects. Historically, major tech shifts (like the internet or electric vehicles) have matured after hype cycles. A correction would likely refocus AI toward real-world productivity and profitability.

What makes the 2025 AI cycle different from past tech booms?
Scale and speed. This AI cycle is global, debt-fueled, and moving faster than any prior tech wave. With trillions invested in chips, data centers, and infrastructure, the potential rewards — and risks — are amplified.

Sources:

1. Nvidia Becomes First $5 Trillion Company - WSJ
2. MIT report: 95% of generative AI pilots at companies are failing | Fortune
3. Big Tech’s $400 Billion AI Spending Spree Just Got Wall Street’s Blessing - WSJ
4. Spending on AI is increasingly fueled by debt, Goldman Sachs says
5. At $1.2 Trillion, More High-Grade Debt Now Tied to AI Than Banks
6. The Creative Dealmaking Behind Meta’s $30 Billion Data Center Financing — The Information
7. Berkeley Lab | 2024-united-states-data-center-energy-usage-report_1.pdf
8. The U.S. has enough copper to meet surging demand from AI data centers. But securing that supply depends on a robust, all-of-the-above strategy.
9. Will the AI Boom Lead to Water and Electricity Shortages? | The Nation

Disclosures:

This communication is general in nature and provided for educational and informational purposes only. It should not be considered or relied upon as legal, tax or investment advice or an investment recommendation, or as a substitute for legal or tax counsel. Any investment products or services named herein are for illustrative purposes only, and should not be considered an offer to buy or sell, or an investment recommendation for, any specific security, strategy or investment product or service. Always consult a qualified professional or your own independent financial professional for personalized advice or investment recommendations tailored to your specific goals, individual situation, and risk tolerance.

Information contained in the materials included is believed to be from reliable sources, but no representations or guarantees are made as to the accuracy or completeness of information.

Dunham & Associates Investment Counsel, Inc. is a Registered Investment Adviser and Broker/Dealer. Member FINRA/SIPC. Advisory services and securities offered through Dunham & Associates Investment Counsel, Inc.