Google has plenty of chips, but has made a strategic decision that changes the rules of the game in the race for artificial general intelligence: it is keeping them for itself. While its competitors are selling off computing capacity at a rapid pace, Alphabet's parent company has prioritized training its own AI models over immediate revenue. It is a move that reveals how seriously major tech companies view AGI not as a future promise, but as a present strategic objective.

The confession that changed everything

The news came to light during Alphabet's second quarter earnings presentation, when analysts from major investment funds directly asked how the company was allocating its TPUs (Tensor Processing Units, its custom chips) between external customers and internal needs. The answer from Sundar Pichai, CEO of Google, was unequivocal: "Our first priority is to make sure that we are allocating what we need to compete at the frontier of AGI development".

It was no slip of the tongue. When another analyst raised the question from a different angle, Pichai was emphatic: AGI comes first, and only then does capacity get distributed among Search, YouTube, its corporate cloud division (Vertex) and other services such as data analytics or cybersecurity.

Cloud business continues to grow (but with caveats)

Google reserving chips for itself does not mean it is selling less. In fact, Google Cloud generated 24.75 billion dollars in the quarter, 82% more than a year ago, with profits of 8.8 billion (a 214% increase). Demand for infrastructure and AI solutions remains voracious, and the backlog of orders reaches 514 billion dollars: these are already signed contracts that customers have yet to consume.

But there is a cost. According to xataka.com, Alphabet's Chief Financial Officer, Anat Ashkenazi, announced that planned investment in AI infrastructure for this fiscal year will rise to between 195 billion and 205 billion dollars, above the previous estimate of 180 to 190 billion dollars. The explicit reason is what Ashkenazi described as a "supply-constrained environment".

Cracks in the strategy

This record spending has brought with it an uncomfortable consequence: Alphabet's free cash flow ended the quarter in negative territory, at -5.9 billion dollars. It is the first time this has happened since 2004. Pichai justifies this investment as a long-term bet to avoid losing major cloud customers, willing to absorb one-off losses in exchange for highly profitable multi-year contracts.

Investors, however, did not see it that way. The company's shares fell 4% at market close following the results. The market, it seems, questions whether the path to AGI is worth burning through cash at this rate.

What does this mean in practice?

Google's decision reflects an uncomfortable truth in the sector: AI chips are the bottleneck. There are not enough to go around, and whoever has them controls the future. While some companies sell capacity as a product, Google is hoarding it as a competitive weapon. It is the end of an era in which the cloud was a commodity: now it is a battlefield.

The question being raised is whether this strategy will work. Will Google achieve a decisive advantage in AGI that justifies bleeding cash? Or will other competitors—with their own chips or preferential access to them—overtake it while it invests?