When the Shovel Seller Starts Funding the Gold Rush

The AI Boom Has a Money Problem

Every gold rush has its shovel seller. While everyone else races to find the gold, the shovel seller makes money from the panic, the optimism and the fear of missing out. In the artificial intelligence boom, Nvidia has become the clearest version of that figure. Its chips power the data centres behind the biggest AI systems, its hardware is treated as essential infrastructure, and its rise has made it one of the most powerful companies in the world.

That part of the story is not difficult to understand. If companies want to build larger AI models, serve more users and run more complex systems, they need enormous computing power. Nvidia has been in the right place at the right time, selling the hardware that makes much of that possible.

The more uncomfortable question is what happens when the shovel seller is not only selling shovels. What happens when it also starts helping fund the miners, the camps, the maps, the roads and the promise that there is gold somewhere ahead?

That is the issue now sitting beneath the AI boom. Nvidia is not just a supplier to the artificial intelligence industry. It is also an investor, partner and financial enabler within the same ecosystem that buys its products. That does not mean the boom is fake, and it does not mean anything illegal has happened. But it does make the story more complicated than the simple version often sold to the public.

The AI boom is no longer only a technology story. It is a financial structure story.

Selling the Future, Then Funding It

The official argument from the AI industry is that demand is overwhelming. Companies say businesses want AI, consumers want AI, developers want AI, governments want AI, and the world therefore needs a vast new layer of data centres, chips, cloud services and power infrastructure. On that reading, Nvidia’s soaring revenues are proof that the future has already arrived.

There is truth in that. AI tools are being used widely, and demand for computing power is real. Millions of people use AI products, businesses are experimenting with them, and the largest technology firms are competing fiercely to build the systems that might define the next decade.

But demand becomes harder to read when the same companies that benefit from the spending are also helping finance it. Nvidia’s investment in OpenAI, its partnership with Anthropic through Microsoft, and its wider investments in AI infrastructure companies and model developers all raise the same basic question: how much of the boom is independent customer demand, and how much is demand supported by money flowing around the same small circle of powerful companies?

This is not a small distinction. A normal customer buying chips with its own money sends one kind of signal. A customer buying chips after receiving investment, support or financing from the chip supplier sends a more complicated signal. The sale may still be real. The revenue may still be booked. The infrastructure may still be built. But the market should be more careful about what that sale proves.

The danger is that investors, governments and the public mistake circular momentum for unstoppable demand.

Nvidia’s Position Is Unique

Nvidia’s position in the AI economy is extraordinary because it sits at the centre of so many layers at once. It designs the chips. It supplies the systems. It works with cloud providers. It partners with AI labs. It invests in start-ups. It helps shape the technical roadmaps of the companies building the models. It is not merely waiting for customers to appear; it is helping create the conditions in which those customers can spend.

That is clever business. It may even be strategically necessary from Nvidia’s point of view. If the world needs more AI infrastructure, Nvidia wants that infrastructure built around Nvidia hardware. If frontier AI labs need capital to scale, Nvidia has an obvious interest in making sure they do not slow down, switch suppliers or build too much of their future around rival chips.

The problem is that this creates a dependency loop. AI labs need Nvidia systems to grow. Nvidia benefits when those labs buy more systems. Investors see Nvidia’s revenue growth and treat it as proof of AI demand. That higher confidence helps justify more investment into AI labs and infrastructure. The labs then spend more on compute, often involving Nvidia hardware.

Again, this does not mean the money is imaginary. It means the structure can become self-reinforcing.

That is exactly why people are starting to ask whether the AI boom is being evaluated clearly, or whether everyone is marking each other’s homework because the short-term gains are too attractive to question.

The OpenAI Example

The OpenAI and Nvidia relationship shows why the issue matters. In 2025, the two companies announced a strategic partnership to deploy at least 10 gigawatts of Nvidia systems for OpenAI’s next-generation AI infrastructure. At the time, Nvidia said it intended to invest up to $100 billion progressively as each gigawatt was deployed.

That was a huge number, even by the standards of the AI boom. It suggested a future in which OpenAI’s growth, Nvidia’s hardware sales and the construction of vast new data centre capacity would be tied together at an unprecedented scale.

By February 2026, OpenAI announced a new $110 billion funding round, including $30 billion from Nvidia. OpenAI also said it had secured next-generation inference and training capacity using Nvidia systems. In simple terms, one of the world’s most important AI companies was raising money from one of the world’s most important AI chip suppliers while also expanding its use of that supplier’s technology.

There may be perfectly legitimate reasons for this. OpenAI needs enormous compute capacity. Nvidia wants to support one of the companies most likely to require it. Strategic partnerships are common in capital-intensive industries. Railways, energy projects, aircraft manufacturing and telecoms have all involved suppliers, customers, financiers and governments working closely together.

But legitimacy does not remove the concern. If OpenAI is funded partly by Nvidia and then uses that capital to secure Nvidia-powered compute, the arrangement deserves scrutiny. Not because it is automatically wrong, but because it blurs the line between market demand and supplier-supported demand.

The public is being asked to believe that the AI future is inevitable. It is reasonable to ask how much of that inevitability is being financed by the companies with the most to gain from it.

The Anthropic Pattern

The pattern is not limited to OpenAI. Microsoft, Nvidia and Anthropic announced a partnership in which Anthropic committed to buy $30 billion of Azure compute capacity, while Nvidia and Microsoft committed up to $10 billion and $5 billion, respectively, into Anthropic.

This is the new shape of the AI economy. The model companies need compute. The cloud companies want to sell compute. The chip companies want their hardware inside that compute. The investors want exposure to the next platform shift. Each party can explain its own logic, but together they form a tightly connected financial ecosystem.

That ecosystem can look impressive when everything is rising. Revenues go up. Valuations rise. Capital expenditure expands. Data centres are planned. New funding rounds are announced. Everyone involved can point to the same thing as evidence of growth.

But a loop is still a loop, even when it is profitable. The question is what happens if end-user revenue, business adoption or productivity gains do not arrive quickly enough to justify the scale of the infrastructure being built. At that point, the issue is no longer whether AI is useful. It is whether the people in charge of the AI boom have overbuilt the future because the short-term incentives rewarded them for doing so.

That is where the scepticism should sit.

Not Anti-AI, But Sceptical of the People Selling It

There is a difference between being sceptical of AI and being sceptical of AI companies. The technology can be useful. It can help with research, writing, coding, accessibility, design, data analysis, customer support, administration and scientific work. Used well, it can save time and open up new possibilities.

The problem is not the existence of the tool. The problem is the behaviour of the people racing to own the tool, monetise the tool, inflate the value of the tool and make the rest of society dependent on the tool before the economics have been properly tested.

Many of the people leading the AI boom are not behaving like careful stewards of a public-changing technology. They are behaving like executives and investors chasing scale, dominance and valuation. They talk about transforming humanity, but the business model underneath often looks much more familiar: raise enormous sums, build as fast as possible, lock in infrastructure, dominate the market, worry about consequences later.

That is the part worth challenging. Not whether AI can help someone write a report or summarise a document, but whether a small group of companies should be allowed to turn a speculative infrastructure race into something the whole economy is expected to absorb.

The technology may have long-term value. The current leadership culture around it often looks worryingly short-term.

The Short-Term Incentive Problem

The AI boom is full of short-term incentives disguised as long-term vision. A start-up raises money at a huge valuation because investors fear missing out. A cloud company commits billions to data centres because it wants to be the platform of the future. A chip supplier invests in customers because those customers will need more chips. Executives announce bigger numbers because bigger numbers create more confidence.

Everyone can claim to be building for the future, but many of the rewards arrive immediately. Share prices rise. Valuations increase. Founders gain influence. Executives gain status. Investors can mark up their holdings. Governments talk about growth. Suppliers book revenue. The story feeds itself.

The risk arrives later.

If AI products do not produce enough reliable revenue, if enterprise adoption is slower than promised, if cheaper models reduce the need for expensive infrastructure, if rival chips become good enough, or if customers simply refuse to pay what the industry needs them to pay, the bill will not remain neatly inside Silicon Valley.

It will spread through markets, pension funds, energy systems, suppliers, public incentives, local planning decisions and workers whose jobs were reorganised around promises that may have been overstated. That is why this matters beyond investors. When a boom becomes large enough, its correction is rarely private.

The gains are often concentrated. The fallout is usually shared.

When the Bubble Question Is Too Simple

Calling the AI boom a bubble may be tempting, but it may also be too simple. Bubbles are often discussed as if they are entirely fake, but history is messier than that. The railway boom built real railways. The dot-com bubble produced real internet infrastructure. Many companies failed, but the underlying technology did change the world.

AI could follow a similar pattern. The technology may be real, useful and important, while the financial structure around it may still become overheated, wasteful and dangerous. Those two things can be true at the same time.

That is why the better question is not simply whether AI is real or fake. The better question is whether the scale of current spending is justified by the value being created now, rather than by the value investors hope might appear later.

The AI industry often asks the public to look at the scale of infrastructure as proof of future demand. But infrastructure can also be a bet. Data centres, chips and power contracts are not evidence that the economics will work. They are evidence that very powerful companies believe, or need others to believe, that the economics will work.

The shovel seller can make money before anyone finds gold. That does not prove the gold rush will pay off for everyone else.

The Public Is Being Asked to Trust the Same People Who Benefit

There is another uncomfortable point here. The people telling the public that AI is inevitable are often the same people who benefit financially if everyone accepts that claim. The companies warning that organisations must adopt AI are selling AI. The chipmakers saying compute demand will keep rising are selling the compute. The cloud companies saying businesses need more AI infrastructure are leasing that infrastructure. The investors talking about the next industrial revolution are already positioned to profit if others believe them.

That does not automatically make them wrong. But it should make the rest of us cautious.

Society has seen this pattern before. A new technology arrives. Its promoters describe it as unavoidable. Critics are dismissed as backward or fearful. Public policy starts bending around the promised future. Workers are told to adapt. Consumers are told to accept change. Local communities are asked to host infrastructure. Energy systems are asked to cope. Regulators are told not to slow innovation.

Then, only later, does everyone ask who made the decisions, who got rich, who carried the risks and whether the benefits were as broad as promised.

That is the concern with the AI boom. It is not that AI has no value. It is that the people in charge of scaling it have every incentive to move faster than society can properly question.

What Real Scrutiny Would Look Like

Real scrutiny would not mean banning AI or pretending the technology is useless. It would mean asking harder questions about the money behind it.

Are AI companies generating enough revenue to justify the infrastructure being built for them? How much of chip demand comes from independent customers, and how much is supported by supplier investment, cloud credits, long-term compute commitments or strategic financing? Are shareholders being shown the full risk of these arrangements? Are regulators keeping pace with deals that make suppliers, customers and investors increasingly hard to separate?

There should also be questions about energy, land, water and public benefit. If huge data centres are being built to serve private AI companies, who pays for grid upgrades? Who gets priority when electricity supply is strained? What happens to local communities asked to host infrastructure? What happens if the boom slows and some of those assets become overbuilt?

These are not anti-technology questions. They are public interest questions.

A technology that could reshape work, energy use, education, media and public services should not be scaled through a financial structure that only experts can understand and only insiders can profit from.

The Gold Rush Needs a Reality Check

Nvidia may still be making one of the most important business bets of the decade. It may be right that AI infrastructure demand will keep growing. OpenAI, Anthropic and other model companies may turn enormous compute spending into products that businesses and consumers genuinely rely on. The shovel seller may end up funding a gold rush that really does find gold.

But there is another possibility. The industry may be building too much, too quickly, on the assumption that future revenue will arrive because everyone involved needs it to arrive. In that version of the story, today’s record sales are not proof of a stable future. They are part of a self-reinforcing race in which the biggest players keep funding the next round because slowing down would make the whole thing look less inevitable.

That is why the circular money question matters. It is not a technical detail for analysts. It goes to the heart of whether the AI boom is being built on genuine demand, strategic fear, financial engineering, or some unstable mixture of all three.

The public does not need to be anti-AI to be sceptical of this. In fact, anyone who wants AI to be useful, durable and responsibly developed should be worried when the companies leading it appear more interested in speed, dominance and valuation than in proving the economics honestly.

Every gold rush creates people who believe the future belongs to them. Some are right. Many are not.

The lesson is simple enough. When the shovel seller starts funding the miners, it is time to stop admiring the size of the boom and start asking who is really paying for it.