The best of times, the worst of times: ARM and Intel in sharp contrast in Q1

Never has the old joke ‘if I were you I wouldn’t start from here’ been so apt in the chip world. The contrast between the fortunes of Intel – announcing its biggest ever job cuts – and ARM, whose quarterly results soundly beat market expectations, highlight the huge advantage of ARM’s starting point in low power and mobile processors, which it is still leveraging despite the challenges some of its licensees are facing because of the slowing of smartphone growth. But that slowdown is nothing compared to the decline of the PC, which of course is Intel’s starting point in device chips. The company eked growth out of its dominance of the PC for longer than many expected, but now has to face the hard facts of the slow death of its most captive market, and its failure to replace that with a strong mobile business.

Intel CEO Brian Krzanich is taking his most decisive action to date, reorganizing the company around infrastructure and the Internet of Things (IoT), which generated $2.2bn of revenue growth in 2015, offsetting PC decline. His firm will “accelerate its evolution from a PC company to one that powers the cloud and billions of smart, connected computing devices,” as he put it.

This sees him looking the end of the PC era squarely in the face, while tacitly acknowledging that mobile devices will not compensate for this. To restore its dominance of device chips, Intel must leapfrog to the next generation of ‘PC replacement’ gadgets, in the IoT, while consolidating its strength in servers by seeing off challenges from other platforms, and extending its own into growth areas like NFV, cloud and carrier network infrastructure.

During its analyst call following the announcement of its first quarter results, Intel CFO said he expects the PC market to decline in “high single digits” throughout 2016, a steeper fall than Intel had anticipated just six months ago, when it was looking for a mid-single digit drop per quarter. IDC and Gartner both say that PC sales fell by between 10% and 12% in the first quarter of this year.

That means Intel needs to decide what will be its best prospect to retain a significant business in device processors, the smartphone having failed, after repeated efforts, to deliver that result. Intel remains a niche player in mobile processors and SoCs, even though its purchase of Infineon’s wireless arm has made it a top five modem supplier. But it was very slow to integrate that technology with its CPUs, highlighting just how far out of its comfort zone it remains in the mobile devices market, compared to expert SoC integrators like Qualcomm.

Now, Krzanich aims to conduct an internal beauty contest, examining all its products and deciding which will be valuable to drive the key platforms in IoT and cloud/infrastructure – and dumping most of the others. When that process is completed, Krzanich said, the company will achieve its highest ever revenue per employee.

That will be partly because Intel will have far fewer employees by then. Its restructuring will involve 12,000 job losses worldwide, about 11% of the workforce and the company’s biggest ever blood-letting (even more than the 10,500 Paul Otellini axed in 2006, at the start of the PC slowdown, though fewer than 18,000 last year at Microsoft). The plan is expected to deliver $750m in savings this year and annual savings of $1.4bn by mid-2017. Intel will take a one-time charge of approximately $1.2bn in the second quarter of this year.

This task of deciding which products should stay or go will fall on the shoulders of Intel’s newest senior executive, former Qualcomm EVP Venkata ‘Murthy’ Renduchintala, who heads the Client and IoT Businesses and Systems Architecture Group. The formation of that operation around Renduchintala – who is seen as Krzanich’s unofficial second-in-command – was the first step in the major reorganization which continued earlier this month with the departure of two veteran business unit leaders, PC chief Kirk Skaugen and Doug Davis, head of the fledgling IoT group. This week, Intel announced that Smith would move to head up sales, operations and manufacturing as soon as a replacement was appointed for his CFO role. That effectively puts a two-person team of deputies under Krzanich, either of whom might be a future CEO.

The newcomer is now charged with evaluating which elements of Intel’s roadmap make sense. “Murthy is doing a complete review of all of our products and will report back to me in the near future and give me a proposal,” Krzanich said.

Some areas of retreat are obvious, such as low end PCs (except Chromebooks, a rare new form factor in which Intel leads ARM). Intel is likely to concentrate on higher margin variations on the PC such as hybrids, gaming PCs and set-top boxes with computing capabilities. However, its dominance is such that, if it exits a particular segment, life will become very hard for manufacturers there, though that could provide a short term boost for AMD.

Other decisions are less clear-cut, but it is not unthinkable that Intel will pull back from the smartphone market where it has spent billions of dollars in R&D, acquisitions and subsidies, for very little revenue and hefty losses. Tablets, in particular, look vulnerable. In a rare misstep, Krzanich decided to subsidize vendors to use Intel chips in tablets last year, which meant Intel topped his sales target of 40m (selling 46m tablet processors), but made heavy losses on the products. Having reversed the subsidy policy, Intel saw its tablet volumes fall by 44% year-on-year in Q116. Analyst Linley Gwennap of Linley Group said: “They still don’t have any marquee wins in smartphones and the tablet business is falling apart. That’s one area they could get out of.”

Krzanich certainly wasn’t hinting at something so drastic, and indeed, the timing would be off in handsets, if not tablets (a sector which is under pressure anyway). But Intel needs to give its boldest handset move yet, its partnerships with Spreadtrum and Rockchip in China, time to work their way into the huge device manufacturing base of that country; and it may finally have won an iPhone slot, though only for the modems (see separate item). But many of the mobile products will surely have to fight hard for their lives in the current climate, including the late-arriving SoFIA SoC and the Atom low power platform itself (Atom has not been updated on the server side for nearly three years; IoT developments increasingly focus on the newer Quark processor; and even in some low power products like Chromebooks, there are Core as well as Atom models).

The CEO is likely to be as unsentimental about mobile products as he is being about PCs. If Intel’s timing in the mobile market has been repeatedly poor, it seems that Krzanich is timing the start of the PC exit rather better. Last year, the sector held up more robustly than expected and buoyed Intel’s performance, but the company was already preparing for the future – 40% of revenues and 60% of profit margin now come from outside the PC space. That still means, of course, that more than half of revenues come from PCs, so now that the writing is clearly on the wall, the firm needs to accelerate change and act  decisively. “It’s time to make this transition and to push the company over all the way to that strategy and that strategic direction,” Krzanich said on the earnings call. “That’s why we wanted to do it now.”

Like Satya Nadella, his counterpart at Microsoft – the other half of the Wintel axis which defined the PC market and is struggling to plan for its death – Krzanich has none of his predecessor’s over-strong attachment to past glories. Like Nadella, his focus is on cloud and the devices which connect to that cloud – his dilemma is which of those devices Intel can realistically dominate.

“We’ve talked about this transformation, that we’re moving from client-centric to a company that’s focused on a much broader set of products, and really focused around the cloud,” Krzanich said. “The cloud, and all the connected devices that connect to that cloud. And that connectivity that brings those devices to the cloud. And that includes the PC, but it’s much more than that.”

The new breed of devices may promise huge volume growth, but they come beset with their own challenges. Most of the unit volume will be in very cheap chips which will not fit well with Intel’s model, and will be better supplied by low cost Chinese manufacturers. “Intel will make money on IoT by making more servers for the cloud not smart lightbulbs,” Gwennap said. Krzanich, who has hung his hat on the IoT since taking the CEOship and invested heavily in wearables, said Intel would be selective about the type of devices it would target, focusing on higher margin areas like automotive, industrial and retail. These will be markets where devices will tend to feature heavy duty connectivity, and the processing power to support machine learning, and so will deliver “PC-like margins”, while also driving more investment in cloud and network infrastructure. “Over the long term this area will be mid-teens growth with, for the most part, PC-like margins and we will try to play to segments that play to our strengths,” Krzanich said.

But with many uncertainties over the IoT model, and Intel’s other growth area, memory, so cyclical, it will be left to infrastructure to deliver the bulk of the turnaround, and a significant growth area will be mobile operators, as they transition to cloud platforms to run their own virtualized networks and to support services for their customers. This is certainly the area where analysts are most positive about Krzanich’s plan. “They are more in position to do a top-to-bottom rack architecture than anyone, and they have a more unified approach than IBM,” commented analyst Nathan Brookwood of Insight64. IBM’s Power architecture, as well as ARM designs, are being trialled by Google and other hyperscale infrastructure players, but despite the signs of new competition, Krzanich says data center products’ average selling prices will rise throughout this year, across the range, from high end Xeon Phi processors to $10 network chips. But he did not dismiss ARM and IBM.  “Having been raised by Andy Grove, I’m always paranoid about the competition,” he said.

By contrast, ARM outperformed analyst expectations in its first quarter results, and is showing signs of success at counterbalancing the slowdown in smartphone growth, with its own shift towards new form factors, especially in the IoT.

CEO Simon Segars said in his statement: “Everyday devices are increasingly being improved by first becoming digital, and then smart, and then connected. This is generating huge amounts of data that needs to be protected, transmitted, managed and stored across the internet. These trends are creating fantastic opportunities for ARM and our partners. They are driving our licensing, as more companies need access to smart processors to build intelligence into more products, and they will drive future royalty revenue as more consumers and enterprises choose to buy smarter and more connected products.”

ARM reports its figures in both US dollars and sterling. In Q116, its dollar revenue was up 14% year-on-year to $398m (a rise of 22% in sterling), and its pre-tax profits were up 14% to £137.5m. Processor licensing was up by 24% to $191.9m.

Although the IP licensor has a completely different model to Intel’s, its results are an important indicator for the whole ARM-based processor sector, and for key Intel competitors such as Qualcomm and Broadcom.

Success across this whole industry will go to the companies which are sharpest at spotting new opportunities and quickest at transferring resources away from slowing markets. Here, ARM is doing better than many of its customers at developing new offerings for high growth segments – 15 of its 16 new licensees during the quarter were for non-mobile platforms, and of 4.1bn ARM-based chips which shipped in the period (up 10% year-on-year), 55% were for non-mobile markets. This indicates a company which is skilfully compensating for the slowing growth of its core revenue stream by diversifying into a number of new areas, while focusing on the higher margin elements of its traditional mobile base (the two-pronged response to slowdown which Intel is now following).

So ARM claimed its average royalty per mobile device is rising despite the price and volume pressure in the smartphone and tablet spaces, because high end designs like 64-bit ARMv8-A, octacore processors and Mali GPUs are gaining increasing penetration in mobile gadgets. It also saw a 10% rise in shipments for network infrastructure, and a 20% rise in microcontrollers and smartcards as – again like Intel – it looks to expand towards both ends of the cloud/IoT product chain.

ARM said that it signed 39 processor licences with 27 companies in Q1, 16 of these first-time customers. Most of the licences were for the three key Cortex families – eight were for Cortex-A, for smartphones and infrastructure; five for Cortex-R for real time embedded applications and automotive; and by far the largest number, 22, for Cortex-M, which targets microcontrollers for the IoT.

But that growth comes at a price – it is the Cortex-A line, especially as it penetrates higher end equipment, which delivers the bulk of the margin.

The UK-based company says its designs now power more than 85% of mobile computing devices, 25% of microcontrollers, 60% of wireless connectivity chips, 15% of networking infrastructure and 7% of automotive (including 95% of in-car infotainment systems).

Q1 2016 – Revenue Analysis Revenue ($m)*** Revenue (£m)
Q1 2016 Q1 2015 % Change Q1 2016 Q1 2015 % Change
Technology Licensing
Processors 135.3 109.3 24% 92.1 71.0 30%
Physical IP 13.0 23.9 -46% 8.7 15.6 -44%
Total Technology Licensing 148.3 133.2 11% 100.8 86.6 16%
Technology Royalty
Processors 191.9 167.5 15% 135.5 110.1 23%
Physical IP 23.8 17.2 39% 16.9 11.4 48%
Total Technology Royalty 215.7 184.7 17% 152.4 121.5 25%
Software and Tools 19.6 14.7 34% 13.6 9.6 42%
Services 14.4 15.6 -7% 9.6 9.8 -2%
Total Revenue 398.0 348.2 14% 276.4 227.5 22%
*** Dollar revenues are based on the Group’s actual dollar invoicing, where applicable, and use the rate of exchange applicable on the date of the transaction for invoicing in currencies other than dollars. Over 95% of invoicing is in dollars.

Intel’s first quarter results:

In the first quarter of 2016, Intel turned in worse than expected figures even though its profit of $2bn was up by 3% year-on-year. Revenues rose 7% to $13.7bn.

Revenue for the Client Computing Group, which covers PC and mobile chips, was $7.5bn, up 2% year-on-year. Revenue for the Data Center Group was $4bn, up 9%.

The chip giant forecast that its second quarter would see revenue fall slightly to $13.5bn. It drew down forecasts of 2016 revenue slightly to mid-single digit growth.

It plans to spend $9.5bn in capex this year, weighted towards the final quarter, when it will be preparing for its first 10nm production capability and to ramp up 3D NAND flash memory production in its Dalian, China fab. “Those two items – 10nm and Dalian – neither are impacted by the restructuring,” Smith said.

Qualcomm’s fiscal Q2 figures:

It is a familiar pattern for Qualcomm to report results in line with market hopes, but see its stock marked down because of a cautious forecast. In its fiscal second quarter, the shares also fell because of reports that Qualcomm has lost some of its Apple business and will see some of the next generation of iPhones powered by Intel modems.

However, the company had some positives to announce – a string of licensing deals in China, which suggests that its settlement last year with the country’s antitrust authority is having the desired effect; and high interest in new advanced Snapdragon models, which promises to help offset weakening demand for smartphones and price competition in chips for low end and midrange handsets.

“The China market is continuing to be strong,” president Derek Aberle said after the report. “We have a more bullish view than we had a few months ago. That’s being offset to some extent by a weakness in some of the other emerging regions, primarily based on some of the macro economic issues we’ve seen.”

However, Qualcomm’s forecast for fiscal Q3 slightly disappointed Wall Street – it is expecting revenue between $5.2bn and $6bn and profit of 90 cents to $1 per share.i

For Q216, it reported net income up 11% year-on-year to $1.16bn, or 78 cents a share, on sales which fell 19% to $5.55bn but were ahead of market forecasts of $5.33bn.

Qualcomm’s revenue has fallen by more than 10% for four quarters in a row as competition rises and growth in core businesses slows – and like Intel, Qualcomm faces the urgent need to accelerate diversification away from the business which made it great, and into higher growth areas like IoT and infrastructure.