APREX · EST. 2026
Demand for Alphabet's AI solutions among businesses and consumers has reached a level that exceeds the company's ability to deliver. That is the backdrop for what now appears to be one of the most aggressive infrastructure offensives in the history of technology.
Alphabet is treating AI infrastructure as a permanent, structural operating cost – not a temporary phase.
The Largest Stock Offering in History
In June 2026, Alphabet completed its largest stock offering ever, raising $84.75 billion, according to TechCrunch. In February of the same year, the company raised a further $20 billion through bonds earmarked for AI infrastructure. Taken together, this represents a financing round of a scale that is rare in the business world.
The purpose is to meet an investment need that is growing rapidly: Alphabet's capital expenditures for 2026 are estimated at between $175 billion and $185 billion, up from $91.4 billion in 2025. The funds are primarily intended to scale up AI computing capacity for Google DeepMind and Google Cloud.
Google Cloud Grows Strongly
The investments appear to be paying off. Google Cloud's order backlog nearly doubled to $462 billion in the first quarter of 2026, driven by demand for AI services in the enterprise market and sales of the company's own TPU chips (Tensor Processing Units), according to the research reviewed for this article.
Industry observers describe Alphabet's approach as an acknowledgment that investing in AI infrastructure is not a one-time event, but an ongoing and structural cost for companies that wish to operate at the technological frontier.
NVIDIA Dominates – But Alphabet Diversifies
NVIDIA still controls approximately 70–80 percent of the market for AI accelerators by revenue as of 2026, and its data center revenues reached $193.7 billion in fiscal year 2026. Alphabet maintains a close partnership with NVIDIA and was among the first to adopt the Blackwell platform on Google Cloud.
At the same time, Alphabet is seeking to spread risk across its supply chain. A notable move is a large order placed with Intel for the production of more than three million TPU chips, with delivery planned for 2028. This marks a departure from previous practice, in which Google's own chips were manufactured by TSMC. Analysts note that the capacity of Taiwanese manufacturers has been pushed to its limits by the AI wave, and that Intel may be able to absorb some of the excess demand.
AMD Bids for Market Share
AMD currently represents a modest challenger with around 5–7 percent of the AI GPU market by revenue, but the company is growing. Its MI300 series of accelerators is attracting attention, in part because MI300X chips are priced 30–50 percent lower than NVIDIA's H100. AMD has set a target of reaching 13–15 percent market share by 2030 and has already signed agreements with major cloud providers and OpenAI.
The overarching trend among large cloud players is clear: a growing desire to spread supplier dependency and avoid being entirely reliant on a single chip vendor. Alphabet's move with the Intel order is a clear signal in that direction.
Risks and Caveats
It is worth noting that projections for AI investment at this scale carry significant uncertainty. Figures for future capex budgets and market shares are analysts' estimates and companies' own guidance – not final accounts. The claim that demand is outstripping supply is Alphabet's own characterization, and the company has obvious incentives to communicate strong growth to the investment market.
Nevertheless, the direction is clear: Alphabet is betting big, and its competitors are watching closely.