Google’s custom AI processors are contributing to record cloud growth, but expanding hardware sales is creating new inventory costs and cash-flow pressures.
Alphabet’s growing artificial intelligence chip business is helping drive record revenue at Google Cloud, but the company’s expansion into hardware sales is creating additional financial pressures as it challenges Nvidia’s position in AI computing.
In its third-quarter financial results, released October 28, Alphabet reported that Google Cloud revenue increased 103% year over year to $30.7 billion, supported by demand for enterprise AI infrastructure, AI services and conventional cloud products.
The company said sales of its custom Tensor Processing Unit (TPU) systems made a meaningful contribution to that growth.
However, the expansion requires substantial upfront investment in processors, networking equipment and data-center infrastructure.
Google’s AI Chip Business Is Expanding Beyond Its Cloud
Google originally developed TPUs to accelerate its internal machine-learning workloads.
The company is now expanding their availability to external customers as organizations seek alternatives to Nvidia’s graphics processing units.
According to CRN’s October 29 reporting, Google Cloud CEO Thomas Kurian said the company has more than 100 significant TPU customers.
The company is also extending its distribution through partnerships with Oracle and AI infrastructure provider Crux AI.
Kurian emphasized that TPU systems combine processors with system software, compilers, networking and other infrastructure needed for model training and inference.
This strategy positions Google as a supplier of integrated AI computing systems rather than simply another semiconductor vendor.
Hardware Expansion Creates Cash-Flow Pressure
The growth is introducing financial challenges.
According to Reuters, Alphabet must purchase and assemble significant amounts of hardware before customers complete payments.
Chief financial officer Anat Ashkenazi explained that unsold TPU systems are increasingly contributing to inventory balances.
Unlike conventional cloud services, selling complete hardware systems can require substantial inventory spending ahead of revenue collection.
Alphabet’s financial disclosures showed approximately $51.4 billion in third-quarter capital expenditures and negative free cash flow of about $9.1 billion.
Those expenditures include broader infrastructure investments and should not be attributed entirely to TPU sales.
Following the earnings announcement, Alphabet shares fell nearly 11% on October 29 as investors assessed rising costs and the company’s outlook.
Nvidia Faces a Different Kind of Competition
Google’s strategy is also attracting large AI developers.
The company previously announced plans to supply Anthropic with TPU capacity worth tens of billions of dollars.
Kurian expects TPU system sales to grow further during the current quarter and accelerate significantly in 2027.
However, Alphabet does not separately disclose TPU revenue or profit margins, making it difficult to determine how the hardware business compares financially with Nvidia.
The distinction matters because high demand for AI computing does not automatically establish attractive margins after accounting for manufacturing, deployment and infrastructure expenses.
The Bigger Question for AI Infrastructure
For AI startups and engineering teams, increasing competition between custom processors and conventional GPUs could create additional deployment options.
However, selecting computing infrastructure requires evaluating more than hardware specifications.
Developers must consider model compatibility, software tooling, networking performance, total operating costs and the difficulty of migrating workloads between computing environments.
For investors, Google’s expansion demonstrates that AI infrastructure competition increasingly involves entire technology platforms rather than individual chips.
The decisive question is whether Google’s integrated computing systems can deliver competitive performance and sustainable financial returns as demand for AI infrastructure continues to rise.
