Late in 2017, Intel (INTC +0.30%) announced that it would be entering the market for high-performance stand-alone graphics processors.
Although Intel has some key go-to-market advantages that could help it gain a reasonable amount of share, such as broad and deep relationships with all major computer vendors and a dominant position in the computer processor market, the reality is that for Intel's stand-alone graphics processors to succeed, they need to be competitive in terms of performance and power efficiency.
Image source: Intel.
A critical enabler of both of those factors is the choice of chip manufacturing technology. Intel historically had a lead over other contract chip manufacturers, such as Taiwan Semiconductor Manufacturing Company (TSM 0.60%), in moving to next-generation manufacturing technologies. However, in recent years, Intel's execution with respect to chip manufacturing technologies has degraded and it's now arguable that the best contract chip manufacturers are ahead of Intel in terms of manufacturing technology.
If Intel cannot maintain parity with the top contract chip manufacturers, then its graphics processors will be at a severe disadvantage to those made by market leader NVIDIA (NVDA 4.43%), since NVIDIA works very closely with TSMC. So, to mitigate potential manufacturing-related risks with respect to its graphics processor efforts, Intel might want to consider tapping TSMC to build these chips.
The benefits
Here's why Intel should tap TSMC to build its stand-alone graphics processors. First, Intel would be guaranteed to have roughly the same length of time from product conception to sale as NVIDIA.
Intel would, of course, still need to design competitive graphics processors using that technology (NVIDIA is the best in the industry at its craft and even coming close won't be easy for Intel to pull off), but at least it wouldn't be held back by manufacturing-related concerns.
TSMC has many years of experience of building high-performance graphics processors for multiple vendors, so Intel would be in good hands here. Manufacturing technology quality, suitability, and time-to-market would not be of any concern to Intel were it to work with TSMC.
The disadvantages
There are, of course, some negatives to Intel using TSMC to build its stand-alone graphics processors rather than building them itself.
First, TSMC doesn't work for free. Intel would need to pay not only TSMC's raw manufacturing costs, but also whatever profit margin TSMC demanded on top of that. TSMC's corporate gross profit margin typically hovers in the range of 50%. If Intel's repeated claims of having a manufacturing cost advantage over the contract chip manufacturers are true, then Intel would likely enjoy better profit margins building the graphics processors itself.
Next, while TSMC is a contract chip manufacturer that works with anyone who will pay, the reality is that NVIDIA is a far bigger and more reliable customer of TSMC's than Intel is. TSMC is likely to work more closely with NVIDIA to tune its manufacturing technology to meet NVIDIA's needs than it is with Intel. This could give NVIDIA a slight, but potentially meaningful leg up in the marketplace.
Additionally, in the case of allocating highly in-demand leading-edge manufacturing technology, Intel is likely to be far lower in the priority queue than NVIDIA.
The ideal situation
Ultimately, the ideal scenario for Intel would be for its manufacturing group to clean up its act and deliver new manufacturing technologies that are finely tuned to Intel's graphics processor product requirements.
However, if Intel can't get its manufacturing act together, tapping TSMC to build its stand-alone graphics processors would be a viable option.
