Intel's path forward: the bull case for a US-government-backed foundry play
Aug 26, 2025
Key Points
- Intel's foundry strategy now hinges on U.S. government equity stake and coercive pressure on tech giants like Apple and Nvidia to place orders, with tariff leverage as the enforcement mechanism.
- The government would need to lock in $40 billion in customer commitments over 18–24 months to justify the Ohio fab investment, functioning as an end-user buyer to offset Intel's process disadvantage versus TSMC and Samsung.
- Government ownership solves Intel's credibility problem by committing to keep the foundry business alive through downturns, a long-term trade-off Intel rejected for decades in favor of margin expansion.
Summary
Intel's foundry bet—manufacturing chips designed by other companies rather than only making its own—has been a thesis since at least 2013, when Ben Thompson first proposed it as a strategic pivot. Intel missed mobile entirely and leaned on cloud computing to stay profitable through the 2010s. The AI boom exposed the core problem: Intel is far behind in GPU leadership, cannot afford to build cutting-edge fabs without $40 billion in customer demand to justify the Ohio facility, and has no natural customers willing to bet their designs on the company.
The bull case for Intel now depends on government intervention. The U.S. government would take a stake in Intel (committing to the survival of its foundry business), fund the Ohio fab build, raise tariffs with strategic carveouts, and pressure tech giants like Apple and Nvidia to place orders. Trump could use a 15% export tax on Nvidia as leverage and threaten larger tariffs. Nvidia could theoretically commission Intel to manufacture its Grace CPU or license designs to Intel. Semiconductor equipment makers like KLA, Applied Materials, and Lam Research could face pressure to invest in exchange for approved licenses.
This is politically fraught. Intel would be prioritizing government interests over fiduciary duty to shareholders. But as one analysis notes, had Intel made that trade-off decades ago—valuing the long term over short-term margin expansion—it would not be in this position now. The mechanics differ from pure subsidy. Unlike the failed 1980s DRAM reshoring effort, the government could function as an end-user buyer in the way the DoD buys rare-earth magnets from MP Materials at a fixed price, building inventory while private capital pursues higher-margin sales elsewhere. If demand for advanced chips is genuinely strong, the government's own purchases could be offset by commercial sales.
Thompson's key insight: without an equity stake, the U.S. has no credible way to promise that Intel's foundry business will survive downturns or competitive pressure. Ownership is the commitment device. Whether it works depends entirely on whether the administration can lock in $40 billion in customer commitments over the next 18–24 months and whether those customers can accept sub-optimal yields and process maturity compared to TSMC or Samsung.