Is AI in a bubble (Part 2)?

AI is moving from experiment… to essential.

Every major industry is integrating it.
Every major company is investing in it.

By late 2025, AI was already an $800B market — growing at a pace that could push it well beyond $1 trillion in the years ahead.

Cloud infrastructure is scaling fast.
AI-enabled devices are multiplying.
Automation is becoming standard.

But here’s the real question…

When trillions flow into this transformation — which stocks stand to benefit most?

Our new report reveals 10 AI stocks positioned across the backbone of this shift — from the companies powering the infrastructure… to those embedding intelligence into everyday systems.

If you want exposure to one of the defining growth trends of this decade, start here.

Download the Report Now

In the first part of this newsletter, we explored the unprecedented scale of capital flowing into AI infrastructure. The numbers are staggering. Big Tech is expected to spend nearly $725 billion on AI infrastructure in 2026, with almost 94% of operating cash flows being reinvested into data centres, GPUs and compute capacity. On paper, the logic appears compelling—AI is growing rapidly, data generation continues to explode, and the companies leading this race are among the most profitable businesses ever created.

But investing has never been about identifying great technologies alone. It has always been about matching expectations with reality.

And this is precisely where the AI story becomes significantly more nuanced.

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

One of the most common arguments supporting today's AI spending is surprisingly intuitive.

Yes, the industry is spending enormous sums today. But as enterprises increasingly integrate AI into their operations, revenues will eventually catch up. Businesses will happily pay for AI because it improves productivity, reduces costs and automates repetitive work.

At first glance, this seems like a perfectly reasonable assumption.

After all, if AI can make every business even marginally more efficient, wouldn't every enterprise eventually become a customer?

That argument sounds convincing until one begins examining the available evidence.

Several industry studies paint a far less optimistic picture.

According to McKinsey, nearly 73% of enterprise AI deployments are failing to achieve their projected return on investment.

Boston Consulting Group (BCG) reports that only around 5% of companies are currently generating substantial financial returns from AI implementations.

Research from MIT suggests an even more uncomfortable statistic—a 95% failure rate in producing measurable financial returns from enterprise AI projects.

Perhaps the most revealing statistic of all is that only 29% of executives surveyed were even able to measure AI's return on investment.

Think about that for a moment.

Companies across the world are spending billions on AI initiatives, yet most cannot confidently quantify whether those investments are actually creating economic value.

This is not an argument against AI.

It is an argument against assuming that every dollar invested today will automatically translate into future profits.

The discussion becomes much more tangible when we look at a real-world example.

A real company.

A real founder.

A real financial decision.

Flo Crivello, founder of an AI startup called Lindy, shared an observation that quickly attracted attention across the technology industry.

Despite operating a relatively small company with approximately twenty-five employees, Lindy's largest operating expense was no longer salaries.

It was AI.

Specifically, the company was spending more on Anthropic's Claude API than on its own payroll.

That single statement reveals something profound about today's AI ecosystem.

Most discussions around AI focus on productivity gains.

Very few focus on the cost of generating intelligence.

When those costs become excessive, businesses behave exactly as economics predicts.

They look for substitutes.

And that is exactly what happened.

Instead of continuing to purchase expensive compute from Anthropic, Lindy shifted its workloads to DeepSeek, reportedly reducing inference costs by nearly 90%.

The implications extend well beyond one startup.

If customers can obtain similar outcomes at one-tenth the cost, pricing power across the industry immediately comes into question.

This is precisely the opposite of what today's infrastructure investment assumes.

Another compelling example comes from Uber.

According to comments made publicly by Uber's CTO, the company exhausted its entire annual AI budget in just four months.

Again, the issue wasn't whether AI delivered value.

The issue was affordability.

Every enterprise has finite budgets.

No matter how transformative a technology may be, management teams eventually ask the same question:

Does the incremental value justify the incremental cost?

That question becomes particularly important when AI providers continue investing billions while simultaneously hoping customers will absorb rising usage costs.

History suggests customers rarely behave that way.

They optimise.

They negotiate.

They search for alternatives.

Exactly as Lindy did.

One of the underlying assumptions behind current AI valuations is that companies such as OpenAI and Anthropic will eventually enjoy extraordinary pricing power.

If every enterprise depends upon their models, then revenue should continue growing rapidly.

But software markets rarely remain monopolistic for long.

As more capable open-source models emerge and cheaper alternatives become available, customers naturally migrate toward lower-cost solutions—especially when performance differences become marginal.

This dynamic has existed across virtually every technology cycle.

Ironically, that could benefit society enormously while simultaneously disappointing investors who funded today's infrastructure build-out.

One of the strongest criticisms comes from Alex Karp, CEO of Palantir.

Palantir occupies a unique position within enterprise software.

Its customers include governments, intelligence agencies and some of the world's largest corporations.

When Karp speaks about enterprise adoption, he is describing conversations taking place inside organisations actually writing the cheques.

Karp has argued that many enterprises increasingly feel they are paying for AI tokens that generate little economic value.

In his view, some AI providers have significantly oversold what current models can realistically deliver.

Whether one agrees entirely or not, the criticism highlights an important distinction.

There is a difference between technological capability and economic value creation.

Even exceptional technology must ultimately justify its cost.

Up to this point, the discussion has focused largely on software.

But the AI boom is also reshaping physical supply chains in ways most consumers never notice.

The global memory chip industry provides an excellent example.

Three companies dominate advanced memory production:

Traditionally, these manufacturers supplied memory chips to companies building laptops, smartphones, gaming consoles and consumer electronics.

Then AI arrived.

AI servers require specialised high-bandwidth memory (HBM), which commands significantly higher margins than conventional DRAM used in consumer devices.

Faced with a simple economic decision, manufacturers redirected production.

Capital always flows toward the highest return.

Industry estimates suggest that roughly 93% of new memory production has shifted toward AI applications.

From a shareholder perspective, this made perfect sense.

From a consumer perspective, it produced unintended consequences.

As production shifted toward AI infrastructure, supply available for traditional consumer electronics tightened dramatically.

The numbers are remarkable.

Dell reportedly observed that the price of one gigabyte of DRAM increased from approximately $0.43 to $2.39 within six months.

This is not financial speculation.

This is real inflation inside a manufacturing supply chain.

Suddenly, Apple's earlier decision begins to make much more sense.

The company wasn't merely responding to temporary shortages.

It was responding to a structural reallocation of semiconductor production toward AI infrastructure.

Consumers effectively began subsidising the AI boom through higher hardware prices.

One could almost describe this as an AI tax—not legislated by governments, but created through market forces.

At this stage, an obvious question emerges.

If enterprise returns remain uncertain...

If AI costs remain high...

If memory prices are creating inflation...

The answer lies in one of the most fascinating concepts in investing:

The Capital Cycle.

The capital cycle has repeated itself throughout financial history.

The sequence rarely changes.

Stage One: High Returns

Stage Two: Overinvestment

Stage Three: Excess Capacity

Stage Four: Survival

History repeatedly demonstrates that revolutionary technologies often survive while the companies that first overbuilt them do not.

History offers a compelling comparison with the telecommunications boom of the late 1990s.

Following the Telecommunications Act of 1996, investors became convinced that internet traffic would grow almost indefinitely.

The assumption was largely correct.

Internet usage did explode.

But expectations became even more extreme.

Companies raced to build fibre-optic infrastructure across the United States.

Over $500 billion was invested in cables, switching equipment and telecommunications networks.

Valuations became detached from reality.

Global Crossing achieved a market value approaching $47 billion despite never generating sustainable profits.

Corvis reached tens of billions in valuation despite essentially having no meaningful revenue.

Investors believed bandwidth demand would justify every kilometre of fibre being laid underground.

They were right about demand.

They were wrong about timing.

By the early 2000s, only 2.7% of installed fibre capacity was actually being utilised.

More than 95% remained idle.

Bandwidth prices collapsed.

Bankruptcies followed.

Roughly $2 trillion of market value disappeared.

And yet, something fascinating happened.

The fibre itself never disappeared.

Years later, YouTube emerged.

Netflix emerged.

Cloud computing emerged.

Smartphones arrived.

Eventually, society needed precisely the infrastructure that had once appeared excessive.

The technology won.

Many investors didn't.

And that may be the most important lesson for today's AI boom.

In Part 3, we'll tackle the biggest question of all: Is AI actually in a bubble?

We'll explore why today's environment is similar to—but also very different from—the dot-com era, what investors like Ray Dalio, Michael Burry and Jeff Bezos are warning about, the two possible paths ahead for AI, and the key lessons long-term investors should take away from this extraordinary technological boom.

Read more about AI here

Join millions of traders putting their knowledge to work on real-world events—from inflation to elections. Buy “Yes” or “No” shares and earn if you’re right.

No house. Peer-to-peer. Cash out anytime.

Get a free $10 to start. Claim it and start trading now.

Start Predicting Now

_{Trade responsibly.}

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.