Intel (INTC -1.41%), the world's largest manufacturer of x86 CPUs for PCs and data centers, is closely scrutinized by tech and financial pundits. But despite all that attention, there are still plenty of misconceptions about its core business. Let's examine four issues that are frequently misunderstood.
1. "Moore's Law" ended years ago
Back in 1965, Intel's co-founder Gordon Moore predicted the number of transistors within the same area of silicon would double every year. That prediction, known was "Moore's Law", drove Intel to continuously develop smaller and denser chips. In 1975, Moore expanded the cycle from one to two years. That revised "law" held up until the past decade.
Using Moore's Law as its foundation, Intel deployed a "tick-tock" cycle for chip upgrades: Each "tick" produced a smaller chip with a higher density, while each "tock" upgraded its architecture but didn't reduce its size. Up until 2012, Intel's "ticks" and "tocks" alternated like clockwork in two-year cycles.
But that cycle gradually lengthened due to the difficulties of manufacturing smaller and denser chips. Intel's latest 10nm chips, for example, were finally launched in 2019 after several delays -- but its next generation of 7nm chips won't arrive until 2023. Moore's Law clearly ended years ago, but Intel continues to tweak its definition, while referring to it as a "metronome for the modern world".
2. Node sizes and chip densities
Intel's decelerating development of smaller chips caused it to fall behind TSMC (TSM -1.43%) and Samsung in the "process race" to create smaller chips.
TSMC and Samsung both started mass producing 5nm chips last year, which seemingly puts both foundries two generations ahead of Intel. However, a simple comparison of node sizes doesn't take into account a chip's density, or the number of transistors within the same area.
TSMC, Samsung, and Intel are the only three foundries in the world that can manufacture chips smaller than 10nm. However, Intel's chips are actually denser than TSMC and Samsung's chips.
Therefore, Intel claims its 10nm chips are actually comparable to TSMC's 7nm chips -- which only puts it one chip generation behind TSMC and Samsung. But that argument might be rendered moot soon, since TSMC will likely start mass producing 3nm chips before Intel launches its 7nm chips.
3. The in-house myth
Intel manufactures most of its chips at its own foundry. That makes it an IDM (integrated device manufacturer) instead of a "fabless" chipmaker that outsources its production to a third-party foundry like TSMC.
When Intel started to fall behind TSMC and Samsung in the process race, many investors and analysts urged Intel to jettison its foundry business and becoming a fabless chipmaker like AMD (AMD -3.23%), which spun off its own foundry business more than a decade ago.
Yet Intel has quietly outsourced the production of its chips to other foundries in the past. TSMC manufactured its Atom CPUs, which mainly powered netbooks, mobile devices, and vehicle infotainment systems, and it recently agreed to manufacture Intel's new discrete GPUs and upcoming 3nm CPUs. Samsung also recently started to manufacture Intel's southbridge chipsets.
Therefore, the notion that Intel is a completely "in-house" IDM is false. In fact, it's become increasingly dependent on TSMC and Samsung even as it doubles down on expanding its own internal foundries.
4. Its "new" GPU business
Intel recently launched a new line of discrete GPUs that target Nvidia (NVDA -1.51%) and AMD's high-end cards. Some investors might assume these GPUs represent a new business for Intel -- but the chipmaker has actually made GPUs for years.
In fact, Intel is the world's largest maker of GPUs, since it bundles low-end integrated graphics chips with its mobile CPUs. If we include those integrated chips, Intel controlled 68% of the GPU market in the first quarter of 2021, according to JPR, while NVIDIA and AMD split the remaining 32%.
Intel tried to launch a high-end discrete GPU over a decade ago, but it abandoned those efforts due to the chip's technical shortcomings. Therefore, Intel's latest DG-series GPUs represent a return to this high-end market -- which is dominated by Nvidia and AMD -- but they're not the company's first GPUs.
How could these misconceptions affect investors?
Reviewing these four points gives investors a clearer understanding of the company's struggles and ambitions. The death of Moore's Law, which Intel still refuses to acknowledge, indicates it's becoming increasingly difficult to manufacture smaller and denser chips.
Its loss of the process lead to TSMC and Samsung -- and its desperate attempts to keep up by producing denser chips and quietly outsourcing its production to those same rivals -- indicate its foundry business will face plenty of challenges over the next few years. Its "new" GPU business also doesn't signify an innovative new step into a new market -- it's merely rehashing an old idea.
I'm not too optimistic about Intel's near-term future, but I think that bullish and bearish investors alike would do well to learn a lot more about the chipmaker by digging deeper into these four points.