Intel (INTC 1.05%) is one of the largest chipmakers in the world and a bellwether of the semiconductor industry. Most investors probably know it's the largest producer of x86 CPUs for PCs and servers, but that it has also struggled with research-and-development blunders and production delays in recent years.
Those missteps caused it to fall behind Taiwan Semiconductor Manufacturing (TSM 0.57%), or TSMC, in the process race to manufacture smaller, denser, and more power-efficient chips. They also allowed Advanced Micro Devices (AMD 0.06%), its rival that outsources its production to TSMC, to pull ahead with more-advanced chips.
Intel is trying to catch up to TSMC again by expanding its plants and upgrading its technology, but the critics believe its 11th-hour effort will fail and result in an existential crisis for the company. The bulls believe government subsidies, especially the pending CHIPS Act, could save Intel.
I've reviewed those arguments in the past, but today I'll focus on three lesser-known facts that changed Intel's past and could impact its future.
1. Intel once developed ARM-based chips
Intel's biggest blunder was arguably its failure to extend its lead in PC and data center CPUs to the mobile-chip market. It lost that market because x86 CPUs consumed too much power for smaller mobile devices -- which prioritized low power consumption over raw processing power.
That shifting market demand led to the rise of ARM, the UK-based chip designer that was eventually purchased by SoftBank (SFTB.Y -1.36%) in 2016. ARM licensed its power-efficient designs to a wide range of chipmakers, including Qualcomm and Apple, which led to its conquest of the smartphone market. Today, more than 95% of premium smartphones are powered by ARM-based processors.
Intel briefly tried to challenge ARM with its lower-powered x86 Atom CPUs for mobile devices, but those chips failed to gain any meaningful traction -- even after Intel paid original equipment manufacturers billions of dollars to try out its chips.
All of that pain could have been avoided if Intel hadn't made two crucial mistakes. First, it turned down a chance to develop a CPU for Apple's first iPhone, believing that Apple could never sell enough handsets to justify the costs of developing a new mobile chip.
Second, Intel failed to expand xScale, the ARM-based chipmaker that grew from its acquisition of the StrongARM team from DEC in 1998. With the proper investments, Intel could have turned xScale into a formidable ARM chipmaker like Qualcomm. Instead, it sold xScale to Marvell in 2006 and focused on the Atom instead. Intel is still suffering from the repercussions of those two mistakes today.
2. Intel's abandoning the memory chip market
Intel is widely known as a CPU maker, but it's also produced memory chips for decades. Back in 1985, Intel abandoned the DRAM market amid intense competition from Japanese chipmakers.
However, Intel went on to develop flash memory chips in 1988. In more recent years, it partnered with Micron Technology (MU 4.34%), one of the world's top memory chipmakers, to co-develop a new generation of 3D XPoint memory chips for high-end solid-state drives.
But in 2020, Intel announced it would sell its NAND business to the South Korean memory chipmaker SK Hynix for $9 billion. Intel will retain the Optane segment, which houses its smaller 3D XPoint business.
The sale will occur in several phases and close in the first quarter of 2025. Until then, Intel will exclude its NAND business from its non-GAAP (generally accepted accounting principles) revenue and earnings.
Intel likely retreated from the NAND market because it was a distant underdog compared to market leaders like Samsung and SK Hynix, and divesting the business would free up more cash for the expansion of its core CPU business.
3. Intel outsources its chips to TSMC
Intel has made a lot of noise about catching up to TSMC by 2025 and bringing back more chip manufacturing to the United States, but it actually outsources production of some of its own chips to TSMC's foundries.
Intel already outsourced the production of its discrete GPUs to TSMC last year. It also intends to outsource some of its upcoming 3 nm and 5 nm CPUs, which will likely arrive in 2023, to TSMC. Intel will still manufacture most of its own chips, but this odd partnership reveals that it can't possibly turn around its business without relying on its biggest rival.
What do these facts tell us about Intel's future?
Intel's mobile blunders and its exit from the memory chip market reveal just how difficult it's been for the company to consistently expand beyond its core market of x86 CPUs. That's also why it's spinning off Mobileye, the automotive chipmaker it acquired in 2017, in an upcoming initial public offering.
CEO Pat Gelsinger, who took the helm last February, realizes that the aging chipmaker needs to double down on manufacturing to stay relevant. But its outsourcing deals with TSMC -- which it made even as Gelsinger publicly claimed Taiwan was "not a stable place" for manufacturing chips -- reveal just how far behind it has already fallen, and how crucial the next few years will be.