Still Digging: Nvidia’s Record Quarter and the $1.3 Trillion Bet Europe Isn’t Making
In a nutshell
everything on the web starts with the domain
For weeks the question hanging over the technology world was whether the AI boom had finally run out of road. Chip stocks had wobbled, investors had started asking the awkward question — where are the returns? — and Nvidia, the company at the dead centre of it all, had just suffered its longest losing streak since 2022. Then, after the bell on Wednesday, Nvidia answered. The boom is not slowing. It is accelerating. And the scale of what comes next should make everyone, especially in Europe, sit up.
What Nvidia Reported
The numbers are difficult to overstate. Nvidia reported quarterly revenue of 96.2 billion dollars, up 106 percent from a year earlier, comfortably beating expectations, with earnings per share more than doubling and a gross margin of 75 percent. Its data-centre business alone brought in 89 billion dollars, up 117 percent, and now accounts for 92 percent of the entire company. Revenue from the big hyperscalers more than doubled to 48.7 billion. Then Nvidia raised its guidance, forecasting 108 billion for the current quarter, and its chief executive projected 70 percent revenue growth for the next fiscal year — far above what analysts had dared to estimate.
This is a company selling, in the well-worn phrase, the picks and shovels of the AI gold rush, and it is selling more of them than ever. Alongside the results came confirmation of the demand behind them: Amazon and Nvidia announced that Amazon Web Services will buy two million Nvidia GPUs. The buyers, in other words, are not slowing down.
The $1.3 Trillion Signal
If one figure from Wednesday deserves to be remembered, it is this. Nvidia's finance chief told investors that capital spending among the top five hyperscalers is expected to reach 1.3 trillion dollars next year, up from 800 billion in 2026. Read that number slowly. More than a trillion dollars, in a single year, from just five companies, poured into the infrastructure of artificial intelligence — data centres, chips, power, cooling. It is one of the largest concentrated capital deployments in the history of private enterprise, and it is happening now, at a pace that is increasing, not levelling off.
That single figure reframes everything else we have reported this month. The memory shortage driving up the price of ordinary devices, the water and power strains of new data centres, the trillion-dollar valuations — they are all downstream of this: a handful of American companies making an all-in wager on an AI future, and buying the means to build it faster than the world can supply the parts.
Boom or Bubble?
The honest question, the one Nvidia's blowout does not settle, is whether this is a sustainable boom or a magnificent bubble. Both can look identical from inside. The case that it is healthy is strong: the revenue is real, the customers are real, the demand is genuine, and unlike the dot-com era, the companies spending the money are among the most profitable on earth. The case for caution is equally serious. Nvidia's extraordinary growth depends entirely on those five customers continuing to spend at a trillion-dollar clip — and the question they still cannot fully answer is when their own vast AI investments will produce returns to justify the outlay. There is circularity in the system, too: Nvidia invests in the very firms, from OpenAI to others, that then buy its chips. History offers a sobering pattern. The railways and the early internet were both real, transformative, and financial bubbles at the same time; the infrastructure survived, but many of the investors who funded it did not. Nvidia selling every shovel it can make tells us the miners are still digging furiously. It does not tell us whether they will strike gold.
Claim and Counter-Claim
The optimistic reading is that Wednesday put the bubble talk to rest. Demand is accelerating, the most sophisticated companies in the world are increasing rather than trimming their bets, and Nvidia's margins show this is not desperate spending but a genuine platform shift with real economic weight behind it. If AI delivers even a fraction of the productivity its backers expect, a trillion dollars a year will look prescient.
The skeptical reading is that concentration is itself the risk. An entire industry's health now rests on five companies' capital-expenditure decisions, and those companies are spending faster than they are earning returns on the last round. When 92 percent of Nvidia's revenue comes from one category, and that category depends on a handful of buyers sustaining historically unprecedented spending, the system is as fragile as it is powerful — a single change of nerve among the hyperscalers would cascade. The honest synthesis: the boom is real and the bubble risk is real, and anyone claiming certainty in either direction is selling something. What is not in doubt is the scale, and the concentration, of the bet.
The European Perspective
Now look at that trillion-dollar table and count the European seats. There are almost none. Nvidia is American. The five hyperscalers making the 1.3-trillion-dollar bet are American, with the partial exception of players who are still not European. The single most consequential capital allocation of our era — the one that will decide who owns the infrastructure of intelligence for a generation — is being made almost entirely without Europe. This is the uncomfortable subtext beneath every regulatory victory we cover. Europe can shape how AI behaves through the Brussels Effect; it wrote the rules that reshaped Claude and constrained ChatGPT's ads. But it does not build the engines. Its flagship response, a thirty-billion-euro gigafactory programme, is real and welcome — and it is roughly two percent of what five American companies will spend in a single year. That is not a criticism of the effort; it is a measurement of the gap. The deeper question this raises is one Europe has been reluctant to confront directly. Regulation is power, but it is the power to set terms on someone else's creation. If the entire physical and economic foundation of artificial intelligence is owned by a handful of firms on another continent, then Europe's sovereignty over its own digital future is, at best, conditional — a tenant's rights, however well-drafted the lease. Wednesday's numbers were a reminder that while Europe debates the rules of the house, the house is being built, at unimaginable expense, by others. The question for the next decade is not whether Europe can regulate AI. It plainly can. It is whether regulating what you do not build, and cannot afford to build, is enough to call yourself sovereign at all.
We are not first. We are right.