As expected, Apple is to buy the majority of Intel’s modem business, bolstering its quest for self-sufficiency in core technologies. This follows an increasingly familiar pattern in which Apple destroys a formerly essential supplier’s product by withdrawing support, only to launch its own version, sometimes by buying up remains of its partner’s business.
Imagination Technologies, Dialog, Samsung and others have lost revenues and killed product lines in the past when Apple decided to go inhouse for components such as the graphics processor and central processor. In the case of Dialog, it bought the firm’s battery management assets and expertise in a $600m deal last year.
Until now, it has remained reliant on third parties for the modem, however, which means it has not been able to develop a fully integrated system-on-chip (SoC) in which the apps processor and modem (and potentially other elements) are combined to save cost, power and footprint.
One result of acquiring Intel’s modem technology may be to enable an iPhone SoC to improve cost competitiveness. Another, of course, will be to reduce or end its dependence on Qualcomm, restored to its former position as Apple’s exclusive modem supplier after a development with Intel failed to deliver the cutting edge 5G chip Apple wanted for its upcoming 5G iPhone. That forced the iDevice maker to settle its legal actions against Qualcomm, ending its hopes of undermining the chip giant’s famous licensing business model at the same time as breaking free from its modem clutches.
The loss of the Apple 5G deal led Intel to announce its immediate exit from the 5G modem market and raised large question marks over the future of the rest of this business, which it acquired from Infineon in 2011, and which is the world’s third largest modem supplier after Qualcomm and MediaTek (though more weighted than Qualcomm towards low end chips).
Apple will pay about $1bn for the Intel unit and will take on about 2,200 Intel employees, as well as some valuable patents. The iPhone maker, which has been an aggressive patent litigator in recent years, generally comes off worse when these fights involve fundamental cellular technologies, as seen in its losses to Nokia and then Qualcomm. However, it is building up its arsenal, and now has 17,000 wireless patents with which to reduce the fees it pays to major IPR owners like Qualcomm.
It is still not clear how much, if any, of the non-5G business Apple will keep going, and whether Intel will carry on supplying 3G and 4G products, or seek to sell or shut down the whole division now it has no prospect of a share of the key growth area in 5G smartphones. It said, when it quit the 5G business, that it would continue to develop modems for non-smartphone devices such as PCs, cars and IoT gadgets, and the structure of the deal with Apple gives it this option in future.
Far more importantly, the deal “preserves Intel’s access to critical IP we have developed” in the course of creating a 5G modem which had, until the Apple blow, been seen as a viable challenger to Qualcomm’s habitual monopoly of any new network generation in its early stages. That IP will be important to Intel’s real opportunity in 5G, network infrastructure chips. As CEO Bob Swan told analysts: “It enables us to focus on the more profitable 5G network opportunity where we are growing and winning share.”
Announcing better-than-expected second quarter results last week, Swan was not focusing on the latest disaster to befall his company’s ill-fated quest to be a strong player in smartphone chips. Instead, he threw the spotlight on the 5G network, which he said will be one of the company’s biggest areas of investment in the coming years.
He told analysts that Snow Ridge, Intel’s 10nm SoC for 5G base stations, will go into production early next year and that two large (but unnamed) telecom equipment manufacturers have already committed to the architecture. Intel is targeting 40% market share in this segment by 2022.
As for Apple, the purchase should enable it to build its own 5G modem and an SoC, though not in time for the first 5G iPhone, expected this year. The settlement agreement announced by Qualcomm and Apple in April included a multi-year chip supply deal. Kevin Krewell, an analyst at chip research firm Tirias, told EETimes that it would take Apple 2-3 years to produce an integrated modem.
The pressure to assert control of the supply chain has been intensified by recent declines in sales of the iPhone, a device family on which Apple is dangerously over-reliant.
Already, the decision to move away from Intel as the 5G iPhone modem partner – reportedly because the prototype chip was not performing to the standards Apple required – may entail delays or design compromises in the first 5G model. Qualcomm will provide a stopgap which is almost certain to support the capabilities Apple wants, but at the cost of getting back into bed with a sworn enemy.
Hence the purchase of Intel’s modem business. Though that unit had not delivered something up to Apple’s demanding standards as yet, it would be easier and cheaper for the iPhone maker to develop its own modem from an existing platform, rather than taking on the gargantuan task of creating one from scratch (the failure of Intel and several other chip companies, such as Nvidia, in this field indicates just how hard it is to challenge Qualcomm in its core market).
But despite the benefits of full control over the design and availability of the modem, and a reduced bill of material cost, going it alone will also expose Apple to the risk of creating an inferior modem to that of Qualcomm. It remains tough to beat the San Diego firm at its strongest game.
According to Reuters, early this year Apple shifted its modem group from its supply chain operations to its hardware design division, headed by Johny Srouji – who used to work at Intel and was hired in 2008 to head up the inhouse SoC developments.
Reports first emerged in April 2018 that Apple was considering developing its own modem, so that it could create a fully integrated SoC. The other members of the top three smartphone club, Samsung and Huawei, both have significant internal modem and SoC products.
Operators are the only area of growth in Intel’s Q2:
The need to focus on operators and network infrastructure was highlighted by Intel’s second quarter results, in which service providers were one of its few growth areas. Overall, it reported revenue down 3% year-on-year to $16.5bn, while its critical business, ‘data-centric’, fell by 7%.
The largest sector within that group is Data Center, in which service providers provided the only growth, of 3%, while cloud fell by 1% and enterprise/government by 31%. Other data-centric businesses are IoT, the Mobileye computer vision platform, memory, and programmable chips (mainly the Altera FPGAs, which are also important to the 5G offering).
Net income was $4.2bn, down 17% from last year.
For the remaining two quarters of the year, Swan is predicting that enterprise and government sales “won’t get dramatically better, cloud will get a little bit stronger, and comms will get a little bit stronger”. Intel forecast $69.5bn in revenue for the full year 2019, a 2% year-on-year fall but a $500m increase from its Q1 guidance.
Of course, the US-Chinese trade wars are a major concern for Intel, both in terms of its ability to sell to China’s huge vendors, and the future of its substantial investments in local companies such as Spreadtrum. “It’s still a little bit unknown about how this China thing will play out, and that’s a big important market for us and that makes me a little more anxious,” Swan told analysts.