Nada Salam

Artificial intelligence is no longer simply a technological contest to build smarter machines. It has become one of the greatest concentrations of capital in modern economic history, with companies barely a decade old commanding valuations that rival the world’s most powerful corporations.

The figures are extraordinary, even by Silicon Valley’s standards.

In March 2026, OpenAI announced a $122 billion funding round that valued the company at $852 billion. Months later, it entered discussions to raise another $30 billion at a targeted pre-money valuation of $1.25 trillion. Yet an October report by The Wall Street Journal indicated that commitments had reached only about $11 billion, suggesting that even the most enthusiastic investors have their limits.

Anthropic, the company behind Claude, has followed a similarly remarkable trajectory. In May, it raised $65 billion at a valuation of $965 billion. By October, it had filed for an initial public offering, with estimates placing its potential market value between $1.5 trillion and $2 trillion.

These are astonishing sums for companies still in the early stages of establishing sustainable business models.

But the numbers also expose a fundamental misunderstanding that increasingly shapes the public conversation about artificial intelligence.

Capital raised is not revenue. Valuation is not cash. And exponential growth is not the same as profitability.

A company valued at $1 trillion does not possess $1 trillion in its bank accounts. That figure represents what investors believe its equity is worth, based largely on expectations of future earnings, market dominance, and the economic opportunities its technology may create.

The distinction is especially important in artificial intelligence, where the financial demands of building the technology are almost as remarkable as the technology itself.

Training advanced models requires enormous computing capacity. Deploying those models to millions of users demands sophisticated data centers, specialized chips, vast amounts of electricity, and some of the most expensive engineering talent in the world.

Unlike earlier generations of software businesses, which could often expand rapidly with relatively modest additional infrastructure, frontier AI companies must continually invest in physical computing capacity.

That creates an unusual economic relationship between the industry’s largest investors and its most ambitious developers.

Technology giants provide financing to AI companies that, in turn, spend substantial amounts purchasing computing services and infrastructure from those same investors.

There is nothing inherently illegitimate about this arrangement. Strategic investment has long played a role in technological development.

But it complicates the interpretation of revenue growth and raises questions about how much demand ultimately comes from independent customers rather than an interconnected financing ecosystem.

Anthropic illustrates another dimension of the challenge.

Despite generating more than $16 billion in revenue during the first half of 2026, the company reported a net loss of $80.7 billion. Most of that loss was attributable to non-cash accounting adjustments associated with financial instruments.

For investors, separating accounting losses from operating cash consumption is essential. Equally important is determining whether the underlying business can eventually generate sufficient cash to finance its expansion without repeated injections of outside capital.

Ultimately, the investment case rests on three assumptions: that demand for AI services will continue growing rapidly, that computing costs will fall as technology improves, and that competition will not erode profit margins before the industry reaches maturity.

None is guaranteed.

Yet dismissing the investment boom as irrational speculation would be equally simplistic.

Artificial intelligence could become an essential layer of global economic infrastructure, comparable in significance to electricity, telecommunications, or the internet. Its applications already extend across software development, scientific research, financial services, healthcare, and industrial operations.

If the technology delivers sustained improvements in productivity, the resulting economic value could be enormous.

The difficulty is determining how much of that value will accrue to the companies developing the models, rather than to the businesses and consumers using them.

History offers a useful distinction.

The internet transformed commerce and communication beyond recognition, but many of the companies financed during the dot-com boom never survived to enjoy its rewards.

A technological revolution can be entirely genuine while the financial expectations surrounding it prove excessive.

That is the central paradox of today’s AI investment frenzy.

The technology may justify extraordinary optimism about the future of the global economy without necessarily justifying every extraordinary valuation attached to it.

For shareholders, the decisive measures will eventually be familiar ones: operating margins, free cash flow, returns on invested capital, and the durability of competitive advantages.

The question is no longer whether artificial intelligence will transform the economy. That transformation is already underway.

The question is whether the profits generated by that transformation will be sufficient to justify the unprecedented sums being committed today.

Financial history does not ultimately judge technological revolutions by the capital they attract, but by the economic value they create once the excitement of financing gives way to the discipline of returns.