Marketdash
Back to News

Google's Homegrown AI Chips Could Be Its Next Blockbuster Business

MarketDash Editorial Team
4 days ago
Wall Street is waking up to the possibility that Alphabet's tensor processing units might transform from an internal tool into a standalone revenue monster worth nearly a trillion dollars.

Alphabet Inc. (GOOGL) has been quietly building something that Wall Street thinks could become its next massive money maker. The company's tensor processing units, originally developed to power Google's own AI ambitions behind the scenes, are now drawing serious attention as a potential standalone business that could reshape the AI chip landscape.

These TPU chips have already helped fuel a 31% rally in Alphabet stock during the fourth quarter, landing it among the top performers in the S&P 500. But here's where it gets interesting: analysts are starting to believe the real payoff comes when Alphabet stops keeping these chips to itself.

The math is compelling. Selling TPUs to third parties could unlock a market opportunity approaching $1 trillion, according to Wall Street estimates. Even if Alphabet keeps them internal, the chips strengthen Google Cloud by dramatically improving efficiency and AI performance, making the cloud business more competitive against rivals.

A Real Alternative to Nvidia

Gil Luria at DA Davidson told Bloomberg that TPUs represent a genuine alternative for companies looking to reduce their dependence on Nvidia Corp. (NVDA). His take? This business could eventually exceed Google Cloud's valuation, which is saying something given that Google Cloud generated over $11 billion in revenue last quarter.

Morgan Stanley's projections add fuel to the thesis. The firm estimates Alphabet could sell five million TPU units in 2027 and seven million in 2028, translating to roughly $13 billion in additional revenue. That's not a rounding error, that's a meaningful new business line.

Alphabet has already started laying the groundwork. The company signed a multibillion-dollar supply agreement with Anthropic PBC and has reportedly entered discussions with Meta Platforms Inc. (META) about TPU access. These moves suggest Alphabet is testing whether there's appetite for its chips beyond its own data centers.

Meta's Move Sparks Debate

James E. Thorne at Wellington-Altus Private Wealth offers a more skeptical interpretation of Meta's reported interest in Google's TPUs. He argues this reflects tactical maneuvering around Nvidia's supply constraints rather than a fundamental shift in the competitive landscape.

Thorne's view is that hyperscalers are treating TPUs as a cost-effective hedge because Nvidia's Blackwell and Rubin graphics processing units face extended wait times. But high switching costs and the friction of moving away from Nvidia's CUDA software ecosystem prevent any wholesale industry migration. In his assessment, TPUs provide supplemental capacity rather than serving as true replacements, and the recent Nvidia sell-off simply represents predictable volatility in an overheated market.

Custom Silicon Gains Momentum

Dylan Patel, founder of SemiAnalysis, sees the competitive dynamics differently. He argues that both Google and Amazon.com Inc. (AMZN) are ramping up custom AI chip production in ways that genuinely pressure Nvidia's position.

Patel notes that Google's TPUs are currently running at full capacity. His argument is that selling them directly to customers, rather than limiting access to cloud rental arrangements, could unlock enormous market value and emerge as Nvidia's most significant long-term competitive threat.

The future power balance, according to Patel, depends on where AI development concentrates. When a handful of tech giants dominate AI workloads, custom silicon designed for specific tasks gains an edge. But if demand spreads across a broader range of industries and applications, Nvidia's general-purpose GPUs maintain their advantage thanks to flexibility and the established software ecosystem.

What's clear is that Alphabet has built something valuable. Whether TPUs become a standalone revenue machine approaching Google Cloud's scale, or remain a strategic advantage that makes Google's cloud offerings more compelling, Wall Street is paying attention. The chip business that Alphabet developed for internal needs might just become its next external growth story.

Price Action: GOOGL stock is up 0.65% at $321.70 premarket at last check on Thursday.

Back to News

Google's Homegrown AI Chips Could Be Its Next Blockbuster Business

MarketDash Editorial Team
4 days ago
Wall Street is waking up to the possibility that Alphabet's tensor processing units might transform from an internal tool into a standalone revenue monster worth nearly a trillion dollars.

Alphabet Inc. (GOOGL) has been quietly building something that Wall Street thinks could become its next massive money maker. The company's tensor processing units, originally developed to power Google's own AI ambitions behind the scenes, are now drawing serious attention as a potential standalone business that could reshape the AI chip landscape.

These TPU chips have already helped fuel a 31% rally in Alphabet stock during the fourth quarter, landing it among the top performers in the S&P 500. But here's where it gets interesting: analysts are starting to believe the real payoff comes when Alphabet stops keeping these chips to itself.

The math is compelling. Selling TPUs to third parties could unlock a market opportunity approaching $1 trillion, according to Wall Street estimates. Even if Alphabet keeps them internal, the chips strengthen Google Cloud by dramatically improving efficiency and AI performance, making the cloud business more competitive against rivals.

A Real Alternative to Nvidia

Gil Luria at DA Davidson told Bloomberg that TPUs represent a genuine alternative for companies looking to reduce their dependence on Nvidia Corp. (NVDA). His take? This business could eventually exceed Google Cloud's valuation, which is saying something given that Google Cloud generated over $11 billion in revenue last quarter.

Morgan Stanley's projections add fuel to the thesis. The firm estimates Alphabet could sell five million TPU units in 2027 and seven million in 2028, translating to roughly $13 billion in additional revenue. That's not a rounding error, that's a meaningful new business line.

Alphabet has already started laying the groundwork. The company signed a multibillion-dollar supply agreement with Anthropic PBC and has reportedly entered discussions with Meta Platforms Inc. (META) about TPU access. These moves suggest Alphabet is testing whether there's appetite for its chips beyond its own data centers.

Meta's Move Sparks Debate

James E. Thorne at Wellington-Altus Private Wealth offers a more skeptical interpretation of Meta's reported interest in Google's TPUs. He argues this reflects tactical maneuvering around Nvidia's supply constraints rather than a fundamental shift in the competitive landscape.

Thorne's view is that hyperscalers are treating TPUs as a cost-effective hedge because Nvidia's Blackwell and Rubin graphics processing units face extended wait times. But high switching costs and the friction of moving away from Nvidia's CUDA software ecosystem prevent any wholesale industry migration. In his assessment, TPUs provide supplemental capacity rather than serving as true replacements, and the recent Nvidia sell-off simply represents predictable volatility in an overheated market.

Custom Silicon Gains Momentum

Dylan Patel, founder of SemiAnalysis, sees the competitive dynamics differently. He argues that both Google and Amazon.com Inc. (AMZN) are ramping up custom AI chip production in ways that genuinely pressure Nvidia's position.

Patel notes that Google's TPUs are currently running at full capacity. His argument is that selling them directly to customers, rather than limiting access to cloud rental arrangements, could unlock enormous market value and emerge as Nvidia's most significant long-term competitive threat.

The future power balance, according to Patel, depends on where AI development concentrates. When a handful of tech giants dominate AI workloads, custom silicon designed for specific tasks gains an edge. But if demand spreads across a broader range of industries and applications, Nvidia's general-purpose GPUs maintain their advantage thanks to flexibility and the established software ecosystem.

What's clear is that Alphabet has built something valuable. Whether TPUs become a standalone revenue machine approaching Google Cloud's scale, or remain a strategic advantage that makes Google's cloud offerings more compelling, Wall Street is paying attention. The chip business that Alphabet developed for internal needs might just become its next external growth story.

Price Action: GOOGL stock is up 0.65% at $321.70 premarket at last check on Thursday.