Intel, once considered a titan of semiconductor design and manufacturing, is reportedly considering an extraordinary move – bringing rival AMD into the fold as a potential partner for its foundry business. If the idea proceeds beyond preliminary talks, it would mark a radical departure from decades of bitter rivalry between the two chipmakers, and a telling sign of just how urgently Intel needs to reimagine itself in the era of artificial intelligence and global competition.
What might once have been dismissed as unthinkable – Intel asking its oldest adversary to build chips on its factories – today reflects the tectonic realities of a shifting semiconductor landscape. It also underscores how far Intel has been pushed into a corner by the rise of Taiwan’s TSMC and Samsung, both of which hold an iron grip on the world’s most advanced chip production.
A rivalry stretching back half a century
Intel and AMD are not just competitors; they are historical antagonists whose story has been one of relentless legal skirmishes, technological jousting and bruising commercial battles. Since the 1970s, when AMD began as a secondary supplier licensed to produce Intel microprocessors, the relationship has been defined by tension. For decades, every market triumph by AMD – from sporadic leaps in performance to cost advantage – was fiercely contested by Intel with a mix of innovation, aggressive pricing and, at times, controversial business practices that attracted the attention of regulators.
In the 1990s and early 2000s, the rivalry became personal among loyalists of both firms. Gamers, PC builders and IT departments would pick sides in debates over which brand delivered better performance or reliability. Intel typically enjoyed the lion’s share of the market thanks to its manufacturing muscle and economies of scale, while AMD survived in its shadow by focusing on value and efficiency. On rare occasions, AMD outpaced its larger rival with products like the Athlon or later its Ryzen processors, but Intel almost always responded swiftly and regained dominance.
For Intel to now entertain discussions about making AMD a key client of its foundry business demonstrates both humility and desperation. It also illuminates the broader problem – Intel’s decades-long reliance on its own integrated device model, where it designed and manufactured all of its own processors, has been shattered by the rise of companies that separate those functions. Today, fabless firms like Nvidia, Qualcomm and even AMD itself thrive by outsourcing chip production to foundries, while Intel struggles to transform into the kind of contract manufacturer it has long resisted becoming.
The foundry experiment and its pitfalls
Intel formally announced its ambition to become a major semiconductor foundry operator in early 2021, with the goal of competing directly against TSMC and Samsung by producing chips for third-party customers. Chief executive Pat Gelsinger framed it as a grand turnaround strategy that would restore Intel’s global importance and deliver much-needed national security benefits to the United States. The American government, alarmed by supply chain vulnerabilities and a reliance on Taiwan, strongly endorsed the plan and supported it with incentives under the CHIPS and Science Act.
But reality has been less encouraging. Intel’s foundry business has struggled to win scale and credibility, with only a smattering of publicly known contracts. Apple, a crucial benchmark customer, has not announced any commitment, and many potential clients remain sceptical of Intel’s ability to catch up with TSMC’s cutting-edge capabilities at the 3-nanometre and sub-3-nanometre nodes. The financials reflect that difficulty: Intel Foundry has been operating at a loss, weighing heavily on the company’s balance sheet just as competition in AI accelerators and server chips intensifies.
Bringing AMD into the foundry fold could provide Intel with a much-needed marquee customer while delivering volumes to help offset its enormous fixed costs. For AMD, it could serve as insurance in case TSMC’s capacity becomes constrained or geopolitical risks around Taiwan disrupt its supply. But such a partnership would be fraught with awkwardness, given the companies’ long-standing antagonism and the obvious risks of sharing vital technical details with a historic rival.
The AI boom reshaping priorities
Underlying this development is the transformative impact of artificial intelligence on the global chip market. The rise of generative AI workloads has shifted demand towards specialised accelerators, high-bandwidth memory and energy-efficient processors on the bleeding edge of fabrication. Nvidia has become the poster child for this revolution, commanding valuations once unthinkable and capturing the lion’s share of AI data centre deployments with its GPUs.
Both Intel and AMD are racing to claim relevance in this environment. AMD’s MI300 accelerators have earned attention as a credible alternative to Nvidia in the data centre, while Intel is pushing its Gaudi line of AI chips against long odds. Yet each company also faces constraints – AMD depends almost entirely on Taiwan’s TSMC to manufacture its most advanced designs, while Intel is simultaneously struggling to execute on its ambitious roadmap of new process nodes.
Should AMD agree to place some of its chips in Intel’s fabs, it would open the possibility of both companies hedging against the geopolitical uncertainty of East Asia while rebalancing their bets in the AI race. For the United States, it would also symbolise the type of cooperation Washington has quietly encouraged – prioritising national resilience in critical technologies over the rivalries of the past.
Financial pressures mounting at Intel
Despite its strong brand and enormous historical base of customers, Intel is under intensifying strain. Years of delayed process technology transitions, failed product launches and declining market share in PCs and servers have eroded its profitability. Data centre operators, once Intel’s most loyal client base, are increasingly shifting companies to AMD or Nvidia silicon, lured by better performance per watt at lower cost.
Intel’s stock price reflects this malaise, having lagged substantially behind peers in recent years. Investors have been patient with Gelsinger’s turnaround programme but confidence is not limitless. Every quarter of red ink from the foundry business adds to the pressure. A deal to onboard AMD as a customer would reassure markets that Intel can generate real, high-volume contracts outside of government subsidies or symbolic partnerships. It could also spur others in the semiconductor ecosystem to take Intel more seriously as a contract manufacturer.
Yet the irony will not be lost on long-time observers: Intel, once accused of using its manufacturing advantage to crush AMD, now finds its best hope for salvation may come from doing business with the very company it once tried to marginalise.
The risk of cultural friction
Even if a commercial deal makes sense on paper, executing it could be fraught. Both companies have deeply ingrained cultures shaped by decades of mistrust. Sharing sensitive information about product designs, process performance or yield issues involves enormous risks, and neither side would enter such an arrangement lightly. Given how fiercely AMD guards its intellectual property, it would require ironclad guarantees that Intel would firewall foundry operations from its own product divisions. Whether Intel’s customers – and shareholders – would believe such assurances is an open question.
There is also the practical matter of capacity. Intel is still building out and ramping several advanced fabs in Arizona, Ohio and Europe, and many of these facilities have yet to prove they can deliver in large volumes at yields competitive with the Taiwanese and South Korean giants. For AMD to entrust a portion of its flagship products to a competitor that has stumbled repeatedly in process execution would represent a calculated gamble. If Intel fails to deliver, AMD could suffer significant setbacks in its time-to-market schedules, potentially ceding ground to rivals in the unforgiving AI accelerator market.
A fragile alignment in a fractured world
The semiconductor industry has always been global, but the geopolitical winds of the past five years have forced companies to consider resilience in ways not seen since the Cold War. Washington is determined to reduce dependence on Taiwan for critical chips, while Europe and Japan are investing billions to shore up their own sovereign capabilities. Intel has positioned itself as the American champion of onshore manufacturing, but achieving scale requires customers – and AMD, ironically, may be too compelling to ignore.
For AMD, aligning more closely with Intel presents both risks and opportunity. On one hand, it could diversify supply and enhance its credibility with policymakers eager to see more chip production outside Taiwan. On the other, it risks giving its fiercest competitor a toehold in its lifeblood business. The calculus will ultimately rest on trust, pricing and whether both firms can set aside decades of animosity for the sake of pragmatism.
What comes next
Whether the talks evolve into a tangible agreement is far from guaranteed. Much will depend on whether Intel can demonstrate sufficient progress in its foundry technology at competitive nodes, and whether AMD believes diversifying production outweighs the perils of greater entanglement. The semiconductor industry is no stranger to unlikely alliances, but this would be among the most symbolic – a once-dominant incumbent forced to look to its adversary for validation.
What is certain is that the landscape that once defined the Intel-AMD rivalry has been irrevocably altered. The battlefield is no longer merely the CPU wars of yesteryear, but the sprawling contest of AI, geopolitics and supply chain resilience. In that context, even rivals of half a century may find themselves drawn into reluctant cooperation.
For Intel, it is nothing less than a fight for relevance. For AMD, it is a chance to consolidate its momentum while insulating itself against risks beyond its control. And for the broader industry, it is a reminder that old grudges can pale in comparison to the realities of survival in the most strategic technology market of our time.
