AI’s Unlimited Data Moment

What the mobile-phone price war can teach us about AI—and the stocks financing it

Remember when your cellphone bill required a degree in advanced mathematics?

You paid for minutes. You paid for texts. You paid for a limited amount of data. And if one of your children streamed a movie without Wi-Fi, you braced yourself for the bill.

Then came unlimited.

In 2016, T-Mobile introduced an unlimited talk, text and data plan—and challenged the rest of the wireless industry to follow. Eventually, unlimited data became less of a luxury and more of an expectation. Consumers stopped counting gigabytes and started using their phones for everything. The smartphone did not fail. It became indispensable.

But the economics changed.

As carriers competed on price, customers received more data for less money. Meanwhile, the carriers still had to spend billions on spectrum, towers and network upgrades. Usage exploded, but the price charged for each additional gigabyte moved toward zero.

AI may be approaching its own unlimited-data moment.

When intelligence goes on sale

For the past several years, artificial intelligence has been priced—and valued—as something scarce, expensive and highly differentiated. The leading models commanded premium prices because they were believed to be meaningfully better than the alternatives. Investors assumed that enormous demand would eventually translate into enormous profits.

The demand part may be right. The profit part is less certain.

The Wall Street Journal recently reported that businesses are increasingly mixing and matching models—using expensive, cutting-edge models for the most difficult work while routing routine tasks to cheaper or open-source alternatives. That is putting pressure on leading providers such as OpenAI and Anthropic to reduce prices. In other words, companies are starting to shop for intelligence the way we once shopped for cellphone plans.

Why pay premium prices for every question if a cheaper model can handle most of them perfectly well?

AI may become more valuable to the world at exactly the moment it becomes less profitable for some of the companies selling it and this is the lesson from unlimited data

The mobile-phone price war offers several lessons for AI investors.

First, lower prices can accelerate adoption.

Once people stopped worrying about data limits, they streamed more video, used more apps and spent more of their lives on their phones. The network became more valuable because it became easier and cheaper to use.

AI could follow the same path. If intelligence becomes inexpensive—or simply bundled into the software we already use—we will use much more of it.

That is good news for the technology. But exploding usage does not automatically create exploding profits.

The wireless carriers still had to maintain their networks. They still had to buy spectrum, upgrade equipment and service their debt. Unlimited plans did not make those costs disappear.

AI providers face a similar problem. They must keep buying chips, building data centers, securing electricity and hiring expensive technical talent—even if competition forces them to charge less for every unit of intelligence.

Second, price wars reward scale.

Large wireless carriers could spread their network costs across millions of customers. Smaller providers had a harder time keeping up, and the industry eventually consolidated, including the merger of T-Mobile and Sprint.

AI may consolidate too. The companies with the deepest pockets, lowest computing costs and largest customer bases may be able to withstand years of aggressive pricing. Smaller model providers may be acquired, pushed into specialized niches or disappear altogether.

Third—and most importantly—the company providing the infrastructure may not capture the greatest value.

The wireless network became essential, but much of the economic value migrated to the companies that controlled the device, the operating system, the applications and the customer relationship.

Apple did not need to own the cellular network. It built the ecosystem people did not want to leave.

The same thing could happen in AI.

The foundation model may become the equivalent of wireless data: necessary, widely available and increasingly difficult to charge a premium for.

The real winners may be the companies that use AI inside a trusted product, proprietary dataset or daily workflow. They may not sell intelligence directly. They may simply use cheaper intelligence to make their existing businesses faster, better and more profitable.

The debt does not go “unlimited”

This is where the analogy becomes especially important for investors.

Wireless carriers could lower the price of data, but they could not lower their interest payments simply because competition intensified.

Neither can the AI industry.

Across the AI ecosystem, companies are committing enormous amounts of capital to chips, data centers, power generation and cloud capacity. Some of that spending is being financed through corporate debt. Some sits in data-center projects, leases, joint ventures and long-term purchase agreements.

The largest technology companies have substantial cash flows and strong balance sheets. They may be able to finance the buildout for years.

The greater risk may be farther down the chain: smaller cloud providers, data-center developers, power projects and other businesses whose financial models depend on consistently high demand and pricing.

Debt-service coverage is essentially a test of whether the cash generated by an asset is sufficient to cover its required debt payments.

If the price of AI falls faster than the cost of supplying it, cash flow gets squeezed. Interest payments do not. Lease obligations do not. Depreciation does not. And the power contract does not politely renegotiate itself because tokens got cheaper.

This is how a price war can move from a competitive annoyance to a balance-sheet event.

A data center built on the assumption of premium AI pricing may be less valuable in a world of unlimited intelligence. A project financed around aggressive utilization forecasts may struggle if customers constantly switch to the cheapest available provider. Refinancing becomes more difficult, lenders demand better terms and equity holders absorb the losses first.

The price of intelligence can fall. The debt used to build it cannot.

Does that mean the AI bubble is about to burst? Possibly—but not necessarily in one dramatic event.

A bubble does not require the technology to fail. It only requires expectations to get ahead of the economics. AI may transform the world and still disappoint investors. The mobile phone transformed the world, but not every handset maker, carrier or network investor became a long-term winner.

If an AI price war intensifies, the market may begin separating the true winners from the companies that simply benefited from the excitement. Lower AI prices will also create winners. Businesses that can replace repetitive work, improve customer service, accelerate research or write software more efficiently may see their own margins rise.

So, should investors sell?

This is where I return to one of the core principles of The Edit: we cannot consistently time the market.

Selling everything because AI looks like a bubble requires two nearly perfect decisions. You have to know when to get out—and when to get back in.

Very few investors get both right.

But “do not time the market” does not mean “ignore what is happening.”

It means responding through disciplined portfolio management rather than an all-or-nothing market call.

I would ask five questions:

1. Has my portfolio become overly concentrated?
   A broad market index can still leave you heavily exposed to a small number of technology companies.

2. Which companies are earning money from AI today?
   There is a difference between generating AI revenue and simply announcing AI spending.

3. Who has genuine pricing power?
   A company needs something customers cannot easily replace: proprietary data, trusted relationships, a powerful ecosystem or a deeply embedded workflow.

4. How strong is the balance sheet?
   In a price war, cash flow and manageable debt matter more than a captivating story.

5. Am I paying for perfection?
   Even an extraordinary company can be a disappointing investment if its stock price already assumes everything will go right.

The answer is not necessarily to abandon AI. It is to rebalance, diversify and distinguish between companies building expensive infrastructure, companies selling increasingly commoditized intelligence and companies using cheaper AI to improve their own businesses.

This material is for educational and informational purposes only and does not constitute investment advice. Investors should consider their individual objectives, financial circumstances and risk tolerance before making investment decisions.