Why the Street Is Wrong About Intel — And What Google Sees That Analysts Don't

The bear case writes itself, and we don't dispute the facts. Intel lost the process technology lead to TSMC in 2018 and has been playing catch-up ever since. Revenue declined from $79 billion in 2021 to $52.9 billion in FY2025. The company posted a net loss of $300 million on what was once the most profitable semiconductor franchise in history. Free cash flow is negative $4.9 billion. The dividend was slashed. Pat Gelsinger was replaced as CEO.

The foundry business — Intel's bet-the-company strategy — has consumed over $50 billion in capex across three years with minimal external revenue. TSMC's lead appears insurmountable. AMD and Nvidia have eaten Intel's server CPU share. The PC market is saturated.

Every word of that is true. And every word of it is priced in at 101x forward earnings and a $296 billion market cap that trades below Intel's replacement value of its manufacturing assets.

Google's commitment to future Xeon processors for AI workloads is not a charity donation. Google evaluates silicon suppliers with the most rigorous TCO analysis in the industry. If Google is committing to Intel's next-generation Xeon platform, it means Intel's Granite Rapids and successor Sierra Forest chips are competitive on performance-per-watt and performance-per-dollar against AMD's Turin.

The timing matters. This commitment comes after Intel 18A — the process node that Intel bulls have pinned their turnaround hopes on — has begun producing test wafers with yields that multiple industry sources describe as ahead of schedule. If 18A delivers even 80% of its promised performance improvements, Intel's server products become competitive again for the first time since 2019.

The historical pattern is clear. There is no partial credit. But the early signals on 18A are the most encouraging Intel has produced in years, and Google's Xeon commitment suggests they're seeing the same data.

Intel's market cap is $296 billion. Its fabrication facilities — including the new fabs in Ohio, Arizona, and Germany — have a replacement cost conservatively estimated at $150-180 billion. The Client Computing Group still generates $30+ billion in annual revenue. The CHIPS Act provides $8.5 billion in grants and $11 billion in loans.

Strip out the fab assets at replacement cost and the CHIPS Act funding, and the market is valuing Intel's product businesses — Client Computing, Data Centre, Network and Edge — at roughly $100 billion. That is less than 2x revenue for businesses that generated 20%+ operating margins as recently as 2021.

The analyst target of $47.23 implies roughly flat upside. But that consensus was set before the Google announcement and before the latest 18A yield data. If even two of the thirty-nine Hold-rated analysts upgrade to Buy, the price target moves materially.