Just the other day, a rumor began circulating that China Electronics is "considering a bid" for Marvell's (NASDAQ:MRVL) mobile chip division. Then, shortly thereafter, MKM Partners published an analysis intended to illustrate why Intel (NASDAQ:INTC) buying Marvell's mobile and wireless business would be beneficial to both companies.
However, I'd like to counter that argument to show why such a deal is not only extremely unlikely to happen, but that Intel wouldn't actually gain much from it (though I believe Marvell would be better for it).
Does Marvell have anything Intel needs technologically?
Marvell is a vendor of very low-cost mobile system-on-chip products. Its highest-end modem technology is, in my view, at least a generation behind what Intel is offering with XMM 7260 in terms of transfer speeds. Further, Marvell's applications processors are generally nothing special from a performance perspective, offering similar performance to just about every other value-focused ARM (NASDAQ:ARMH) licensee.
In other words, at the system-on-chip level, just about any technology that Marvell can bring to the table, Intel already has competent in-house efforts for. I expect that when Intel launches its SoFIA 3G/SoFIA LTE chips later this year, this will become more apparent to investors and industry analysts.
This doesn't work from a business perspective, either
As I said before, Marvell plays in the low end of the mobile system-on-chip market. This is a market that Intel stated that it plans to back out of with its own internally designed system-on-chip products. The company also signaled that would instead focus on working with third-parties to integrate key Intel-designed IP into chips aimed at this market segment.
Where, then, would Marvell's chip unit fit into this strategy? Spreadtrum is already a far larger player in the mobile chip market than Marvell is, according to Strategy Analytics, and Marvell doesn't bring anything to the table for higher-end mobile applications processors, where Intel has signaled that it will continue investing organically.
About Marvell wanting to exit this business ...
The curious thing here is just how violently Marvell's stock price moved up upon the mere rumor that it would be exiting the mobile chip business. Even though it is a high revenue growth business, Marvell is likely losing money in the business.
Indeed, when questioned by John Pitzer of Credit Suisse about how close the company is to breakeven, and what kind of market share it would need in order to have a profitable business, Marvell's Sukhi Nagesh had a very interesting response.
It is not appropriate for us to give the breakout at this point due to competitive reasons. And in terms of the scale that will give us the ability to be very profitable on a net operating basis, I'd say somewhere within about 15% to 20% market share in this space.
Think about this one for a moment. Marvell claims that in order to be "very profitable on a net operating basis" (whatever "very profitable" means), it needs to have 15-20% market share of what appears to be the cellular chip market. To put this into perspective, according to Strategy Analytics' various reports, the smartphone applications processor market should be about a $20 billion market in 2014.
Now, in the third quarter of 2014, Strategy Analytics reports that Qualcomm took 51% revenue share in this market, followed by Apple with 14% share, and MediaTek with 13% share. This implies 22% left for everybody else, including Samsung (which has sizable captive demand), Spreadtrum (which has a strong presence in the China market), HiSilicon (which also has sizable captive demand in Huawei handsets), Marvell, and others.
Getting to 15-20% of the total smartphone applications processor market (or even 15-20% of the merchant smartphone applications processor market) seems like a very tall order. While I don't know if the rumors that Marvell is shopping this business around are true or not, I wouldn't be the least bit surprised if it were.
However, if it is, and if it manages to find a buyer, I very seriously doubt that it will be Intel.