Intel’s $500 Billion Question: Where Are Foundry Proof Points?

Intel’s (INTC) stock has climbed more than 5x over the past 12 months, moving from roughly $19 to a recent range near $120 currently. Intel has added close to $500 billion to its market cap – not a small number.

Two forces are driving it.

Resurgent CPU demand, with agentic AI workloads making server processors more central to the AI buildout than markets expected a year ago. The bigger narrative has been optimism around Intel Foundry, the manufacturing arm CEO Lip-Bu Tan is trying to turn into a real external business.

The investment case isn’t hard to understand. AI is driving demand for advanced manufacturing capacity, customers are looking to diversify beyond Taiwan, and Intel is the only U.S. company that both designs and manufactures leading-edge chips domestically. The company is trying to transform its foundry business from an internal cost center into a world-class contract manufacturer.

But the gap between the narrative and the numbers remains wide. External customers contribute only a sliver of foundry revenue, losses remain substantial, and the strategy has shifted multiple times in just two years. If Intel Foundry is becoming the next great semiconductor business, where are the proof points?

What Q1 2026 Actually Shows

Intel Foundry generated $5.4 billion in revenue in Q1 2026, up from $4.7 billion a year earlier. External foundry revenue was $174 million. The operating loss came in at $2.4 billion, roughly flat with the $2.3 billion loss a year ago. For full-year 2025, total foundry revenue was $17.8 billion against an external contribution of just $307 million, with a $10.3 billion operating loss for the year. The foundry remains almost entirely an internal supplier to Intel’s own chip designs. That distinction matters. Manufacturing chips for Intel validates the technology, but the economics of a foundry improve only when external customers trust the process enough to commit meaningful production volumes. See Intel’s segment financials

A Strategy That Kept Shifting

Under Pat Gelsinger, Intel positioned 18A as the flagship node for both internal products and external customers simultaneously. A closer look at Intel’s 18A tech When Lip-Bu Tan took over in 2025, he redirected the external pitch toward the next node, 14A, treating 18A largely as an internal-only process given its early yield struggles. Improved yields and renewed customer interest in 2026 pushed Intel back toward marketing 18A and a new variant, 18A-P, to foundry customers again. These strategic shifts create uncertainty for customers making multi-billion-dollar, multi-year manufacturing decisions, where process stability and roadmap consistency are critical.

Yield Is Improving, Still Trails TSMC

While Intel does not disclose yield figures for its manufacturing nodes, industry estimates suggest Intel 18A Yields are currently in the 50% to 60% range for the Panther Lake compute tile, with further improvements expected as production matures. Yields are critical because they determine how many functional chips are produced from each wafer, directly affecting manufacturing costs, margins, and a foundry’s ability to compete for external customers. Even so, Intel’s yields are likely still below those of a fully mature leading-edge TSMC process, where yields often exceed 70% to 80%, depending on die size.

Who’s Actually Shown Interest

Microsoft (MSFT) has a confirmed custom silicon collaboration with Intel, although the manufacturing node has not been disclosed. AWS is working with Intel on custom Xeon and AI fabric chips, while Apple (AAPL) has reportedly received an early Intel 18A-P design kit for evaluation. Nvidia (NVDA) and SoftBank have also taken equity stakes in Intel, signaling confidence in the company’s broader strategy, though not a commitment to manufacture chips at Intel Foundry. Importantly, there is a significant difference between evaluating a process, securing a design win, and committing production volumes. Receiving a design kit or validating a manufacturing process is an early milestone, but meaningful foundry revenue is generated only when customers commit wafer volumes and move into high-volume production.

The Proof Points Still Needed

The next proof points are straightforward: a major external customer committing meaningful production volumes, sustained high yields at commercial scale, and external revenue becoming a meaningful share of foundry sales. Until those metrics improve, Intel Foundry remains a promising manufacturing platform, but not yet a proven foundry business.

Intel’s foundry story is a compelling one, but a compelling story and a quality investment are not necessarily the same thing. When it comes to protecting and growing wealth over the long term, a disciplined portfolio approach tends to pay better than concentrated single-stock bets. The Trefis High Quality (HQ) Portfolio has consistently outperformed its market benchmark since inception, delivering cumulative returns of over 105%.