Big Tech's AI Spending Reaches $665 Billion: Investment or Bubble?

Big Tech's AI spending could reach $665 billion as early as 2026, but the question that really matters in the markets is no longer how much companies are investing, but when that money will start to come back. After two years in which Wall Street financed almost every project with "AI" in the title, asset managers are shifting their focus from spending figures to the concrete returns expected in the coming quarters.

Capital expenditures related to AI by major tech companies are set to reach between $635 and $665 billion by 2026. This figure should be viewed in a broader context: global investments in artificial intelligence, across all sectors, are expected to exceed $1 trillion by that same year.

Leading this race are hyperscalers Amazon, Microsoft, Alphabet, and Meta, the four entities that are effectively shaping the infrastructure of the entire sector with their budgets. The most recent quarterly reports from Microsoft and Amazon have confirmed that the demand for cloud services remains very strong, to the point that capacity constraints continue to limit the speed at which these companies can grow.

The largest asset managers estimate that these tech giants will collectively add about $340 billion to annual operating cash flow by 2027. This is the number that, according to portfolio managers, makes the otherwise staggering spending figures more digestible: just extend the time horizon and the bet starts to make financial sense.

Richard Clode, an analyst at Janus Henderson, predicts that hyperscalers will grow profits and cash flow at a faster rate than the incremental increase in capital spending by 2028. This is the turning point on which investors are building their return forecasts.

Despite the monstrous figures invested, the stocks of hyperscalers have lost ground compared to the rest of the tech sector: in the most recent earnings season, the Philadelphia Semiconductor Index rose by 75%, while chip manufacturers, those selling the "shovels" in this gold rush, have been rewarded more generously than those actually digging.

The shortage of computing capacity for AI has created an unexpected opportunity for a group of smaller companies: the so-called Neocloud providers, cloud platforms specializing in GPUs, which have benefited simply from the fact that the demand for computing power for AI still exceeds the available supply.

CoreWeave has gained about 50% in recent months, while Nebius has surged over 200%, both riding the wave of AI capacity shortages and the resulting pricing power. The risk for these operators, however, is equally clear: as hyperscalers bring new capacity online, that premium pricing power could quickly evaporate.

Why does this point matter? Because this is where the credibility of the entire spending cycle is at stake. According to wealth managers' calculations, the monetization of AI should grow between 5 and 13 times current levels to justify the capital spending plans already announced by Big Tech. This is not a marginal detail: it is the variable that separates a visionary investment from a spending bubble without adequate returns.

However, AI adoption in business is still in its early stages. Companies are mostly testing pilot projects, not yet fully loaded production workloads. The transition from experimentation to large-scale implementation could unlock the revenue growth necessary to make that $635 billion in capital spending a deal rather than a gamble. It remains to be seen how long this transition will really take, and whether investors will have the patience to wait until 2028 to see profits outpace spending growth.