The balance of power has shifted.

Article author and source: GeekPark

On June 22, 2020, on the virtual stage of Apple’s WWDC, Tim Cook announced a decision that shook the entire PC industry—Macs would abandon Intel and fully transition to Apple’s own chips.

At that moment, a 15-year partnership was sentenced to death.

Six years later, Apple and Intel are once again sitting at the same table. But this time, the roles are completely reversed—instead of Apple using chips designed by Intel, Intel is now manufacturing Apple’s own-designed chips on its behalf.

This story is more captivating than any business textbook.

01 The breakdown of a 15-year marriage

Back in 2005, Jobs personally "killed" PowerPC at WWDC, announcing that Macs would fully embrace Intel. It was Intel’s golden era, with the x86 architecture dominating the entire PC world; Apple’s choice of Intel was almost inevitable.

The honeymoon period of this partnership lasted a long time. Intel helped Apple produce the first-generation MacBook, the first aluminum iMac, and even the iconic “trash can” Mac Pro, still revered by designers today. From 2006 to 2020, every Mac was powered by an Intel chip.

But toward the latter part of the collaboration, cracks began to appear.

Intel's process advancement has slowed significantly, with its 14nm node being refined for four to five years, leading the industry to jokingly refer to it as "14nm+++." This has directly hampered the product strength of Macs—issues such as thermal throttling in the MacBook Pro, the controversial butterfly keyboard, and the awkward positioning of the Touch Bar all stem from Intel chips' high power consumption and lackluster performance improvements.

Apple's most prized "hardware-software integration" experience has been strangled by its suppliers' pace.

Meanwhile, Apple’s ARM chip design capabilities, honed over years on the iPhone and iPad, have matured to an astonishing degree. The A-series chips have seen annual leaps in single-core performance while maintaining extremely low power consumption. Apple began seriously evaluating a bold idea—what if this capability were brought to the Mac?

In November 2020, the M1 chip was released, and the answer was revealed.

The M1 chip in a $999 MacBook Air outperforms the Intel MacBook Pro priced at twice as much, boosts battery life from 10 hours to 18 hours, and eliminates the need for a fan altogether. Industry reviews were nearly unanimous—it wasn’t just an upgrade, it was a generational leap.

Apple completed the full transition in less than three years. In June 2023, the last Mac Pro powered by Intel processors was discontinued, marking the official end of the Intel era.

During this breakup, Cook also left behind a widely quoted remark. It was reported that he privately told TSMC’s founder, Morris Chang, that Intel “just doesn’t know how to run a foundry.”

At the time, this sentence was almost a death sentence for Intel’s foundry dreams.

02 Compound Interest After Six Years

If Apple and TSMC's collaboration proceeds smoothly, this story would likely end here.

But in the world of supply chains, there is no "happily ever after" ending.

Apple has entrusted TSMC with the sole manufacturing of all its advanced-process chips—the A-series chips used in iPhones and the M-series chips used in Macs and iPads. While this has delivered exceptional performance over the past few years, it has also created a significant single-point-of-failure risk.

This risk became a reality in 2026. Cook explicitly acknowledged on this year’s Q1 earnings call that iPhone 17 series shipments were constrained by production capacity—TSMC’s supply of the A19 chip was insufficient. As major AI chip customers like NVIDIA aggressively compete for TSMC’s capacity, Apple found itself waiting in line.

Supply chain diversification has shifted from an optional exercise to a necessity.

Over the past year and a half, Apple and Intel have engaged in intensive negotiations and recently reached a preliminary agreement under which Intel will manufacture some of Apple’s chips. The U.S. government also played a facilitating role, with Secretary of Commerce Lutnick and President Trump directly involved in brokering the deal.

But what truly won Apple's approval wasn't Washington's will—it was Intel's actual deliverables.

03 Cards in Intel's Hand

Back then, Cook criticized Intel for being unable to do contract manufacturing, but today, at least on paper, Intel has changed dramatically.

Since taking over in spring 2025, Intel’s new CEO, Chen Liwu, has made Intel Foundry the cornerstone of the company’s turnaround. His key weapon is the 18A process—a 1.8nm node designed to compete directly with TSMC’s most advanced 2nm technology. This production line has already begun mass manufacturing at Intel’s facility in Arizona, with Intel’s own Panther Lake mobile processors among the first products.

The entry point for Apple's collaboration with Intel is likely the entry-level M-series chip—the one used in the MacBook Air and iPad Pro. These chips have an annual shipment volume of approximately 15 to 20 million units, a significant scale but not involving Apple’s core flagship product lines. Apple has already signed a non-disclosure agreement and obtained Intel’s 18A-P process PDK (process design kit), and internal simulation work is underway. If everything proceeds smoothly, shipments could begin as early as the second half of 2027.

Notably, Apple is more likely to wait for the upgraded version of 18A—the 18A-P. This version offers a broader range of transistor types for flexible customer configurations, delivering approximately 9% higher performance at the same power consumption. Analyst Tim Bajarin believes Apple will most likely wait until the 18A-P matures before commencing full-scale production, with this milestone potentially achievable as early as next year.

Apple's strategy is clear—use entry-level products to test the waters, keep its flagship chips with TSMC, and build its own "second supply chain."

But Intel has secured more than just Apple. NVIDIA has invested $5 billion in Intel and will produce custom data center CPUs on Intel’s manufacturing lines; Elon Musk’s Terafab project (serving Tesla, xAI, and SpaceX) has also chosen Intel for manufacturing; Microsoft and Amazon AWS had already signed contracts earlier. Intel’s stock has risen over 200% this year and is up approximately 433% compared to a year ago.

This company is indeed coming back to life.

04 The exam has not started yet.

But the story won't be this simple.

In the contract manufacturing chip business, there is a brutally cold metric—yield. It refers to the proportion of functional, qualified chips on each wafer, directly determining costs and delivery capability. One of the core reasons TSMC dominates the contract manufacturing market is its exceptionally high yield, far surpassing that of its competitors.

Bajarin, President of Creative Strategies, put it bluntly—both Intel and Samsung must prove they can match TSMC’s yield levels. Apple’s obsession with yield is likely the most extreme in the entire consumer electronics industry. The company has an extremely stringent supplier certification system, and any new manufacturing partner must undergo a lengthy validation process.

Analyst Rob Enderle also noted that Apple executives privately worry that Intel may not yet match TSMC in terms of scale and process node maturity.

In other words, Intel has secured its ticket in, but the most grueling exam—production yield—has just begun.

And there’s also a practical issue: if Intel-manufactured chips are more expensive and have lower yields, that difference must either be absorbed by Apple, squeezing its profit margins, or passed on to consumers, raising device prices. A highly upvoted comment on Hacker News put it plainly—if the goal is simply to avoid making TSMC the sole supplier of advanced chips, then Intel is indeed the most obvious choice, but “obvious” doesn’t mean “cost-free.”

05 The Drama of Role Reversal

Looking back at this entire timeline, the drama is almost overwhelming.

In 2006, Apple partnered with Intel, using Intel-designed chips to build Macs. In 2020, Apple grew frustrated with Intel’s performance and cut ties to develop its own chips. In 2023, the last Intel-based Mac was discontinued, marking the final separation. In 2026, Apple returned—but this time, not with Intel’s chips; instead, Intel will manufacture chips according to Apple’s design.

The party A user has become the party A client; Intel’s role has shifted from “chip designer” to “contract manufacturer.” This is not only a reversal of the relationship between the two companies, but also a microcosm of the broader shift in power structures within the semiconductor industry—where design capability and manufacturing capability are rapidly diverging, and those who control design are increasingly seated at the head of the negotiation table.

Apple never makes meaningless moves. When it switched from Intel x86 to its own ARM architecture over a decade ago, many said it was "impossible"—yet it completed the full transition in less than three years. This time, Apple’s decision to re-engage with Intel is equally driven by careful business strategy.

However, the last breakup changed Intel’s fate—whether this reconciliation can change Intel’s destiny remains to be seen in two years.