Over the last several years, microprocessor giant Intel (INTC +1.05%) has been working hard to bring down the losses associated with its mobile product development. Said losses peaked at just over $4.2 billion in 2014, but Intel says that they came down by nearly $1 billion in 2015 and are expected to come down by at least another $800 million in 2016.
Intel's former Mobile and Communications Group general manager, Hermann Eul, at the launch of the company's Bay Trail platform for tablets. Image source: Intel.
This means that the losses in 2016 should be about $2.4 billion -- still substantial, but far better than where they were in 2014.
The loss reductions have come from a combination of cost-cutting (e.g., cancelling projects and laying off workers) as well as a significant reduction in "contra-revenue" payments as its tablet processor shipments have dried up.
Unfortunately, much of what has allowed Intel to dramatically reduce those losses are pretty much one-off impacts. Significant additional operating expense reductions are unlikely to be feasible considering the previous cuts, and the company's contra-revenue program looks just about done.
This means that the next phase of Intel's mobile loss reductions needs to come from good old-fashioned improved business performance. The first place for Intel to start would be with its cellular modem efforts.
Focusing on the modem
Intel recently won the modem spot inside of Apple's (AAPL +0.00%) latest iPhone 7-series smartphones, splitting the orders with wireless giant Qualcomm (QCOM +0.24%).
This win is likely bringing in some much-needed revenue to the chipmaker's mobile division. And, if those modems are being sold at positive gross profit margins (they likely are), then those sales are likely helping to bring Intel's mobile losses down in 2016 compared to what they were in 2015.
However, it's also likely that the margins on these modems aren't great. Not only does Intel's modem provide inferior performance to its Qualcomm-built version per Cellular Insights (which suggests that Intel can't charge as much as Qualcomm can), but the modem itself is believed to be built by Taiwan Semiconductor Manufacturing Company (TSM +1.40%).
The lack of pricing power coupled with the fact that Intel relies on a third party to manufacture the part (and that third party makes about 50% gross profit margins on those parts) means that Intel is likely pocketing very little in terms of gross profit on these products.
For Intel to improve its mobile profitability, it needs to make some changes to both improve what it can charge for its modems as well as the competitiveness of its products.
What Intel needs to do
To bulk up its modem average selling prices, Intel will need to narrow the gap with Qualcomm's offerings in terms of modem performance and features. Right now, Intel's best is several generations behind what Qualcomm has in the marketplace -- a fact that likely forces Intel to sell these modems on the cheap.
Intel's mobile losses are still substantial and the company has made it clear the investing in modem technology is a key priority for the company. At this point, I don't think Intel's problem is that it's not investing enough in research and development, but rather inefficient use of those research and development dollars.
Management needs to work to identify why its spending on modems is yielding sub-par results and take appropriate action to fix those problems.
Beyond that, Intel isn't doing itself any favors by having these modems built externally when it runs its own chip manufacturing operations (and routinely boasts about how its costs are superior to what its manufacturing competitors can deliver).
Intel is expected to build its next-next generation modem (XMM 7560) on its own 14-nanometer manufacturing technology. The good news is that by moving the modem in-house, Intel benefits in a couple of ways:
- Intel doesn't have to pay a contract chip manufacturer roughly 50% margins; the company can keep that for itself.
- Intel should be able to deliver a substantial power consumption benefit, all else equal, because its 14-nanometer technology is much more efficient than TSMC's old 28-nanometer technology. This could improve Intel's product competitiveness.
The bad news is that a transition from XMM 7480 (built by a third party) to XMM 7560 (built by Intel) won't begin to kick in to the company's financials until mid- to late 2018.
A potential mobile loss trajectory
Intel may provide investors with an expected trajectory for its mobile losses during its investor meeting in February 2017. Ahead of that, I'd like to put my predictions out there.
In 2017, Intel should get a revenue/gross profit benefit from having a full year of modem shipments to Apple relative to 2016, which should help overall profitability. Additionally, if Intel was still shipping significant quantities of mobile products that required contra-revenue payments during 2016 (and the first half of 2016), those should fade significantly during 2017.
I don't think the loss reduction in 2017 will be anywhere near what it was in 2015 or in 2016, but something on the order of $200 million to $300 million might be reasonable to expect.
In 2018, Intel probably won't be able to get any significant benefit from contra-revenue payments abating as I expect that program to be done long before then. During the second half of 2018, Intel could get a product margin/average selling price improvement from the XMM 7560, which could help further drive down those losses.
Without Intel scoring significantly new mobile business outside of Apple, it'll be difficult for the company to meaningfully boost revenue to start driving down the losses.
Intel has indicated that it still intends to remain in the mobile processor game, so if the company starts shipping significant volumes of mobile processors to various smartphone vendors in the 2018 time frame (something I frankly doubt), then those loss reductions could accelerate.
We'll almost certainly know more about Intel's mobile strategy and longer-term financial expectations in a few months.
