Google Gives Intel Foundry Its Biggest Chance in Years

According to The Information, Google has selected Intel to manufacture more than three million of its custom Tensor Processing Units (TPUs) for production in 2028. This is not just a large order — it represents a rare and strategically vital entry for Intel into the supply chain of one of Nvidia’s primary competitors in AI hardware.

For Intel, this deal is a potential game-changer for its long-struggling foundry business. For Google, it is a pragmatic step toward scaling its AI infrastructure while reducing heavy reliance on a single manufacturer. And for the broader industry, it signals the beginning of a more diversified semiconductor supply chain at the most advanced process nodes.

The TSMC Bottleneck

TSMC has effectively become the choke point for the entire AI boom. Demand for high-performance AI accelerators is growing faster than the Taiwanese giant can expand capacity, even with massive investments in new fabs. Leading tech companies — including Nvidia, Apple, Tesla, Amazon, and Microsoft — are actively exploring alternatives to avoid being constrained by one production hub.

Diversification is no longer optional; it is risk management on a multi-billion-dollar scale.

Why This Deal Matters for Google

Google’s TPUs power much of its AI training and inference across Google Cloud and consumer products. While the company has long designed its own chips, manufacturing has largely depended on TSMC.

By placing a major order with Intel (reportedly covering roughly half of its planned TPU output for 2028), Google achieves several goals:

Greater production capacity to meet surging internal and cloud demand.
Reduced geopolitical and capacity risk tied to Taiwan.
Stronger negotiating position with its primary suppliers.
Continued independence from Nvidia’s dominant GPU ecosystem.

Google has reportedly spent months testing Intel’s advanced packaging technologies, including EMIB, and is now confident enough to commit at scale.

Intel’s Make-or-Break Opportunity

Intel Foundry Services (IFS) has struggled for years to attract major external customers despite heavy investment and the U.S. CHIPS Act support. Landing a multi-million-unit order from Google — a hyperscaler with demanding requirements — is the most credible validation yet that Intel’s 18A and future process nodes can compete at the cutting edge.

This deal positions Intel as a potential “second source” for leading AI chips.

Other companies are watching closely:

Nvidia is reportedly evaluating Intel’s 18A process and advanced packaging for future multi-chip GPU designs.
SK hynix is testing Intel’s EMIB for HBM integration.
Additional hyperscalers and chip designers may follow if Intel delivers reliably.

Success here could transform Intel from a struggling IDM into a true foundry powerhouse.

Broader Industry Implications

A stronger Intel Foundry benefits the entire ecosystem. It adds resilience to AI supply chains, encourages competition on process technology and pricing, and supports Western efforts to rebuild semiconductor manufacturing capacity.

No one expects Intel to overtake TSMC overnight. But this Google order marks a meaningful shift from “PowerPoint foundry” to proven alternative. For the market, more advanced-node capacity cannot come soon enough.

The AI race is not just about who builds the best chips — it is increasingly about who can actually manufacture them at scale. Google’s bet on Intel is a smart hedge that could pay dividends for the entire industry.

Google Gives Intel Foundry Its Biggest Chance in Years

The TSMC Bottleneck

Why This Deal Matters for Google

Intel’s Make-or-Break Opportunity

Broader Industry Implications

Subscribe to our newsletter