More woes for Qualcomm as the European Union competition authority suspended its investigation into the US company’s planned acquisition of The Netherlands’ NXP Semiconductors. The process should be restarted once the parties provide some allegedly missing information, but the delay has sparked speculation that there is deeper trouble ahead for the deal – an acquisition which is hugely important to Qualcomm’s future growth and its ability to rise above current pressures including antitrust probes, rising Chinese competition, slowing smartphone growth, and litigation with Apple.
According to Reuters, the delay is because Qualcomm and NXP have not provided sufficient information as requested by the investigators. The regulator had been scheduled to issue its ruling by October 17, but that will now be delayed. Once the companies supply the necessary information, the clock will be restarted and deadline adjusted.
The European Commission has already doused any hopes Qualcomm may have had of getting its acquisition cleared without significant concessions. The EC has expressed concerns that the deal could “lead to higher prices, less choice and reduced innovation in the semiconductor industry”. Although the US Department of Justice has cleared the deal without conditions, the EU was bound to take a more critical stance, given that NXP is one of the major chipmakers that rose out of former European industrial giants (NXP was spun out of Philips, and later acquired the US’s Freescale, itself formerly the chip arm of Motorola).
There are few overlaps between the two companies’ product ranges, so competition should be preserved for chip buyers in Europe. One possible area of sensitivity is, apparently, NXP’s smartcard technology, Mifare, which enables contactless payments and other transactions. Reuters’ unnamed sources said that “rivals had urged the European Commission to ensure they would still be able to use NXP technology known as Mifare once the deal is done”.
Rob Lineback, senior market research analyst at IC Insights, told EETimes there were two possible areas of concern – that Qualcomm would sell Mifare on to a Chinese business “as part of its efforts to win approval from China for the acquisition of NXP”; or that chip firms fear having to pay higher sums for licences to Mifare technology under Qualcomm’s ownership.
Qualcomm will also have to convince the Chinese competition authority, MOFCOM, to clear the NXP deal without significant conditions or delays. China may aim to force Qualcomm to divest certain key technologies or businesses – such as Mifare – in order to strengthen the national policy of building a homegrown semiconductor industry and IPR base.
Such indications may dampen NXP shareholder enthusiasm for the transaction too. Last week, Qualcomm extended its offer period for NXP shares yet again. So far, 12.5% of the total shares have been tendered, and that figure is actually moving downwards (from a peak of 17.2% in March, four months after the bid was launched).
Some reports suggest that NXP activist investors are withdrawing their shares in the hope that Qualcomm will up the price, arguing that it is undervalued by the $110-a-share cash offer, which equates to about $38bn excluding debt. A group of activists, led by Elliott Management, started to push for a higher price in May.
The companies are a long way away from the 80% acceptance level which is required to close the deal. Qualcomm said it will continue to extend the offer each month until that level is reached, or the deal fails.
“We remain on track to close the transaction this calendar year, and integration planning continues across both companies,” said Qualcomm CEO Steve Mollenkopf on the US firm’s earnings call in April, and that line has not changed since, despite the regulatory and shareholder stresses.
Indeed, it is very important that Qualcomm succeeds in its biggest ever takeover. NXP will, overnight, transform its position in one of its key target growth markets, automotive, and will also bring it valuable assets in other important areas, such as security and the overall Internet of Things, where Qualcomm hopes to expand, in order to offset the stagnation of its core smartphone segment.
The combined company would have an addressable market worth $138bn in 2020, with leadership potential in mobile, automotive, IoT, security, RF and networking, according to Qualcomm’s calculations. There are many obstacles even if the deal is approved – a tough consolidation process for a company unused to such large acquisitions; a dramatic change to its current culture and business balance; the challenges of dealing with NXP’s legacy businesses without dragging down the whole performance, and of assimilating a foundry model into a fabless environment.
But all those challenges could prove worthwhile to up Qualcomm’s game in key adjacent markets for its wireless business, saving a great deal of time and R&D expense to attain leadership in segments like auto, where NXP had over 14% share of the $27.4bn sector in 2015.
If the acquisition succeeds, Qualcomm will also gain established positions in microcontrollers and embedded processors in many IoT and industrial areas.
When the deal was announced in Mollenkopf at the end of October last year, he focused on four areas where NXP would strengthen Qualcomm’s offering.
In its core smartphone sector, its new division will bring some relatively commercially marginal but still valuable additions such as NFC (a leadership position via MiFare) and the emerging embedded secure element (eSE) technology.
In automotive, NXP has products across the whole value chain from in-vehicle infotainment (IVI) to security, radar and in-car networks.
In the broader IoT, its strengths lie in its comprehensive microcontroller range plus authentication, transportation systems, secure ID and payments. This reflects the way that Qualcomm will need to surround its modems and SoC products with a far wider range of other technologies even than it has done in handsets, to address the diversity of the IoT.
The fourth area to be highlighted was networking, in which NXP brings embedded communications processors and RF technology, most of it from its Freescale buy, to add to Qualcomm’s 802.11 portfolio, much of that also from an acquisition, of Atheros. This area also includes wireless infrastructure DSPs (digital signal processors), which could get Qualcomm back into the mobile networks market which, apart from its small cell platform, it abandoned two decades ago after settling a legal dispute with Ericsson.
In other words, this will not be a deal about cutting costs or eliminating a rival – it will be all about accelerating Qualcomm’s moves to broaden its offering. The US firm also values its prey’s capabilities in sales and its broad distribution networks.
Most important of all, NXP would help Qualcomm enrich its roadmap towards 5G, when all kinds of networks – cellular, WiFi, IoT, millimeter wave and wireline – will converge. It said it wants NXP to help it “unlock the 5G ecosystem” by driving mobility deep into new markets. The Dutch firm brings a crown jewel in this respect, since its Digital Networks division has been working on a converged platform to support many kinds of connectivity, built around its programmable modem technology – which will fit right into the heart of Qualcomm’s own strategy (though programmable modems are a sore point at the moment, with Nvidia suing Qualcomm and claiming it forced Icera, the UK soft modem pioneer which Nvidia acquired, out of business.)
NXP demonstrated the first implementation of its programmable modem technology at 2016’s 5G World show last June. The company has created a new architecture that allows a fully programmable PHY layer for various wireless standards, such as WiFi 802.11ac and pre-standard 5G. Companies can create different modems for different technologies, using the same basic architecture.
But it also allows customers to future proof their solution as standards change and become refined – they can deploy pre-standard 5G NR now for testing, for example, and then upgrade as standards mature simply by conducting a software upgrade to the installed units.
Until now it has been tough to build a cost-effective programmable solution for mass scale, as Icera discovered. Now, NXP says it has created a solution that is fully comparable to a fixed implementation in terms of power and area.
And there is a broader architecture around this product. In a blog post for RCRWIreless last month (http://www.rcrwireless.com/20170605/opinion/converged-architecture-5G-tag9), Wim Rouwet, a senior principal engineer at NXP Digital Networking, wrote: “While enterprise and service provider communications have been stuck in the rip-and-replace model of periodic network upgrades, a funny thing has happened in the quest for ever-faster speeds. Networks’ physical layers have converged on OFDM and wide channel bands, regardless of media (wired or wireless), band (baseband, sub-6 GHz, or millimeter wave), and standards body (ITU or IEEE). In many cases, these various network technologies will simultaneously offer the same user network access. For example, the user may be served by G.Fast broadband via WiFi, sub-6 GHz LTE, and mmWave-based 5G cellular.”
But of course, many aspects of the network have not converged, and most of them continue to change. As Rouwet wrote, “OFDM parameters, such as subcarrier spacing, vary, as much as a result of the different paths committees take toward standardization as for technical reasons. Massive MIMO makes sense for 28 GHz and higher carrier frequencies owing to their small antennas and poor signal propagation. At Layer 2, networks differ significantly because of differing topologies and usage models. Meanwhile, vendors continually upgrade their designs, and the industry periodically updates its standards.”
He outlined the NXP approach to dealing with all this diversity and change, via a single hardware/software modem architecture which could support multiple modem implementations through firmware loads.
Key components of the architecture would include:
PHY: Single physical layer architecture and implementation across converging wireless and wired standards, providing programmability and flexibility without sacrificing efficiency.
MAC: Tight integration with an optimized packet processing engine that is scalable to support the required multi-Gbps MAC layer throughput.
Packet processing (networking) engine: Seamless integration with networking and control IP designed to support complex packet routing applications, network services, as well as support for remote provisioning and virtualized networks.
SoC: Integrated solution leveraging the key technologies—programmable PHY and MAC, packet engines, ARM cores and associated memory, interconnect and high-speed I/O.
Development: An ever-growing software development ecosystem that OEMs and operators can use to speed their time-to-market with differentiated solutions; initial application areas targeted include 802.11/Wi-Fi (2.4/5 and 60 GHz) and 3GPP (5G) wireless access technologies.