Alphabet's TPU Chips: The Quiet AI Advantage

Alphabet has been designing its own AI chips, called tensor processing units (TPUs), for years. Initially, these chips were used internally for Google's products and cloud customers. This year, Alphabet began selling TPU systems directly to customers for use in their own data centers.

In the second quarter, Alphabet recognized revenue from TPU system sales for the first time. Management noted it's still a small piece, with most revenue expected in 2027. A Citizens analyst estimates around $3 billion in TPU-related revenue this year and $25 billion next year.

This move comes as Google Cloud is gaining momentum, with revenue growing 63% in Q1 and 82% in Q2 to over $24 billion. The cloud backlog stands at $514 billion, with just over half expected to convert to revenue within 24 months.

Alphabet's TPU strategy is not about replacing Nvidia but about creating a second profit stream and gaining more control over costs. The company has agreed to provide credit backstops for third-party data centers and power infrastructure, adding some financial risk.

Despite risks, this is a positive step for Alphabet, which now has a fast-growing cloud unit, a resilient search business, and a new chip business.

The TPU business gives Alphabet two key advantages: it can run its own AI services and those of its cloud customers on hardware it controls, reducing reliance on Nvidia's supply and pricing. Second, it can sell that hardware as a product, creating a new revenue stream.

This is significant because it diversifies Alphabet's revenue beyond advertising, which currently accounts for 68% of total revenue. The TPU business also strengthens Google Cloud's competitive position by offering customers more options.

While TPU sales are still small, the potential is large. The estimated $25 billion in revenue next year would be a meaningful contribution. Moreover, it demonstrates Alphabet's ability to innovate and monetize its AI investments.

The move also positions Alphabet as a player in the AI chip market, albeit a different one from Nvidia. It doesn't need to win the entire market; it just needs to capture a share of the growing demand for AI infrastructure.

Overall, the TPU business reinforces the buy case for Alphabet, showing it can adapt and thrive in the AI era.