Sources: Nvidia tested Intel's 18A to make chips but stopped moving forward; interviews with 20+ people describe how dealmaker Lip-Bu Tan gave Intel a lifeline
It was a Thursday before dawn in Silicon Valley when Intel CEO Lip-Bu Tan found himself under attack by the president of the United States.
Context & Ripple Effects
Intel’s turnaround had already been constrained by internal disagreement over capital raising and an AI acquisition, with the board stalling Tan’s proposed moves. Tan then sought to demonstrate a path for U.S.–Intel cooperation in a planned meeting with Trump as political pressure on the company mounted.
The reported 18A evaluation puts a commercial limit on that recovery narrative: political support may provide Intel breathing room, but a leading prospective customer has not converted its process test into production business. Later coverage of Washington pressing customers to use Intel fabs shows how foundry demand became a policy concern as well as a market one.
First-order effects
- Intel loses the near-term validation and potential foundry volume that an Nvidia 18A production commitment would have delivered.
- Nvidia retains the results of its 18A evaluation without taking on a manufacturing commitment, while Intel and Tan must continue pursuing external-fab customers alongside the company’s lifeline efforts.
Second-order effects
- The non-conversion raises the bar for Intel’s foundry sales effort: prospective customers will look for evidence that process evaluations can become durable production awards.
- U.S. support for Intel is likely to face greater pressure to translate into customer commitments, rather than relying only on ownership or financing measures.
Third-order effects
- If major chip designers continue to test alternative manufacturing nodes without committing volume, second-source capacity will remain strategically valuable but difficult for new foundry challengers to monetize.
- The episode points to a semiconductor market in which industrial policy can sustain domestic capacity, while customer qualification and ecosystem confidence still determine whether that capacity becomes commercially self-supporting.
The trend: Intel’s recovery illustrates the growing separation between strategic support for domestic chip capacity and the harder task of winning recurring leading-edge foundry demand.