Article illustration 1

Intel’s relentless pursuit of Moore’s Law—the foundational principle that transistor counts double every two years—faces its most severe threat yet, not from physics, but from economics. During Intel’s Q2 2025 earnings call, CEO Lip-Bu Tan issued a stark warning: investment in the company’s next-generation **14A (1.4nm-class) semiconductor process node hinges entirely on securing "a meaningful external customer" to share the astronomical costs.

The Economic Reality of Leading-Edge Nodes

"Our external foundry strategy has always been rooted in the economic reality of semiconductor manufacturing," Tan stated. "The increase in capital costs at Intel 14A make it clear that we need both Intel products and a meaningful external customer to drive acceptable returns... I will only invest when I'm confident those returns exist."

This hardline stance formalizes a strategic shift hinted at months earlier by CFO David Zinsner. While Intel’s 18A node progresses with internal products shouldering the financial load, 14A’s development—requiring extreme ultraviolet (EUV) lithography tools like ASML’s High NA systems costing over $350 million each—demands external revenue to justify the investment. Failure to secure a major client could force Intel to abandon 14A entirely.

Moore’s Law at a Crossroads

Intel’s potential retreat from the leading edge marks a profound inflection point. For over half a century, since co-founder Gordon Moore first articulated his observation in 1965, Intel epitomized the relentless drive to shrink transistors and pack more into each chip—from thousands to hundreds of billions. Each node transition faced physical hurdles, but the financial burden now appears insurmountable for a single player. As feature sizes push below 2nm, R&D and tooling costs have skyrocketed, turning Moore’s Law from a technical challenge into a fiscal one.

The Gelsinger Legacy and Tan’s Reckoning

Former CEO Pat Gelsinger’s ambitious IDM 2.0 strategy aimed to transform Intel Foundry Services (IFS) into a third-party powerhouse, spreading node costs across external clients. Four years later, after product cancellations, massive layoffs, and the spin-off of IFS into a separate entity, Gelsinger departed. Tan, his successor, is wielding a ruthless focus on profitability:
- Cancelled the Intel 20A node (intended to bridge 18A and 14A)
- Divested majority stakes in Altera and automotive unit Mobileye
- Abandoned planned fabs in Germany and Poland

Ironically, Intel now increasingly relies on rivals like TSMC—the very foundries Tan needs to outcompete for 14A customers.

What Comes After Moore?

While TSMC and GlobalFoundries continue pursuing sub-2nm nodes, Intel’s potential exit from the bleeding edge underscores an industry-wide reckoning. Past predictions of Moore’s Law’s demise (2009, 2011, 2013) proved premature, but the convergence of quantum effects, EUV complexity, and capital intensity suggests a fundamental shift. If Intel steps back, the mantle of scaling leadership falls to TSMC—but even its success may redefine Moore’s Law as a collaborative endeavor, not a solo race. For developers and hardware engineers, this signals an era where architectural innovation, advanced packaging, and software-hardware co-design gain primacy over pure transistor density. The end of Intel’s solo pursuit doesn’t spell doom for progress, but it irrevocably ends an epoch defined by a single company’s ability to outpace physics through sheer will and investment.

Source: The Register