Analysts with Bernstein Research distributed a note to clients after microprocessor giant Intel (INTC -1.60%) hosted its Feb. 9 investor meeting. In the note, the analysts say that Intel plans to introduce cellular modems manufactured in its 14-nanometer technology (rather than on older, third-party manufacturing technologies) in the second half of 2018.
The analysts say that Intel is not only planning to introduce these chips, but that they plan to ship "tens of millions of units" in the second half of 2018.
Intel wireless chief, Aicha Evans. Image source: Intel.
"Given the only high volume purchaser of stand-alone modems remains [Apple (AAPL -0.81%)], this suggests share at least maintained if not increased," the analysts write.
The report also says that Intel was "coy on CDMA capability" but that the analysts "believe it is a possibility," noting that if Intel does indeed include CDMA capabilities in this modem, it would be a "further potential headwind for Qualcomm (QCOM -2.53%)."
In this column, I'd like to explain why the move to an in-house manufactured modem is so important for Intel and Apple, and why inclusion of CDMA capabilities could negatively impact Qualcomm, as the Bernstein analysts suggest.
The advantages of going in-house
Right now, Intel's modems are manufactured by a third party (believed to be Taiwan Semiconductor Manufacturing Company (TSM -0.55%), though Intel hasn't, to my knowledge, officially confirmed this). Moreover, they are manufactured on relatively old 28-nanometer chip technology.
By moving the modem in-house, Intel should enjoy several benefits compared to its current manufacturing arrangement.
First, Intel must pay TSMC for each wafer of chips that the latter manufactures. This means that Intel not only bears the burden of the raw materials and manufacturing costs, but TSMC marks those wafers up and makes significant profit margin per wafer sold (in the ballpark of 50%).
By moving this to in-house manufacturing, Intel is still on the hook for raw materials and manufacturing cost (and some will argue that Intel's per-wafer manufacturing costs are higher than TSMC's), but it doesn't need to fork over profit margins to TSMC.
Instead, Intel could theoretically be able to either boost its own profit margin on the chips that it sells to Apple (this assumes that Intel's wafer costs aren't so high that they offset not having to pay TSMC's margins), or maintain the same margins that it did before, but lower its chip prices to Apple (potentially opening share gain opportunities).
Beyond that, though, newer manufacturing technologies tend to bring power efficiency benefits over older ones. The transition from TSMC's 28-nanometer technology to Intel's 14-nanometer (presumably an "optimized" 14-nanometer+ or 14-nanometer++ variant) technology should enable a significant reduction in both active and stand-by power -- potentially boosting battery life.
The more technically competent Intel's modems are (better power efficiency is clearly a valuable example of that), the more likely that Apple is going to want to use them in its products.
The CDMA question
Right now, Apple uses Qualcomm modems in iPhone models that support the CDMA wireless standard and Intel modems in the iPhone models that don't. If Apple had to, it could use Qualcomm modems in every single iPhone sold and it wouldn't be a problem; the same can't be said about the Intel modems which wouldn't work properly on CDMA networks.
If Intel can integrate CDMA capability into the modem that it's expected to ship to Apple for the 2018 iPhone, then Apple will presumably be able to use either Intel or Qualcomm modems interchangeably.
That's a negative for Qualcomm because it opens the company up to either additional potential share loss (Apple could keep Qualcomm in the loop but further reduce its allocation), or it could lead to an erosion of Qualcomm's pricing power (in a recent lawsuit, Apple alleges that Qualcomm charges it extra for modems with CDMA capability enabled).
Neither situation would be good for Qualcomm's Apple business.
As to whether we'll see CDMA in Intel's 2018 modem, it's interesting that Intel was reportedly "coy" about this, per Bernstein. Perhaps Intel is working to build the technology into that 2018 modem, but the company isn't confident that the implementation will be robust enough to pass carrier certification in the timeframe that it would need to for Apple to be comfortable using it.
I do expect that Intel will eventually implement CDMA in future modems, though.
