Q2 Special: Telco 2.0 News Review

Posted on

Softbank x ARM; Apple, Samsung, and Huawei Q2s; Q2 shipments

Softbank astonished the world last week by buying ARM Holdings, the British chip designers behind 95% of the world's mobile devices, for $32bn in raw cash. That's "offer you can't refuse" territory, especially as ARM is typical of successful tech companies in that its employees own a lot of the company's shares, and the deal instantly makes millionaires of quite a few of them.

The attraction is pretty obvious - as well as the huge mobile business, ARM's Internet of Things product line is growing fast, and there is more and more interest in using its technology in the data centre and network domains.

Interestingly, Japanese currency traders have been aware for some time that there was an unknown, but very big, buyer in the market for sterling. It turns out Softbank has been accumulating the currency it needs to close the deal for weeks, ever since the big post-referendum sell-off. This points to a couple of things - first of all, the 15% devaluation of sterling made ARM 15% cheaper without affecting its revenue as it prices in dollars, but secondly, as always with Softbank, the clever tricks leave you wondering where Masayoshi Son is going with this.

Softbank promised to increase the company's headcount in the UK, double it abroad, and leave the current management in place - you can see why the deal was attractive, but you might also argue Softbank has paid a silly price for a deal driven by sheer opportunism. And it's no wonder the transaction has become politically controversial - the new UK government had just trailed a new, more interventionist industrial policy when it backflipped and greeted the deal as great news. Will ARM's precious stockpile of patents, for example, stay within ARM Holdings plc or be transferred to some other Softbank company?

Meanwhile, Apple Q2s are in. Paradoxically, although profits for the quarter were down 27% and revenues 15%, the shares rose 6%, as the news was significantly better than expected and in line with Apple's own forecast. The problem can be summed up as "everyone's waiting for the new phone in Q3, and in the meantime, the iPhone SE is taking the strain and that's not as pricey". Services, interestingly, were up by just under a billion - this article makes the case that the company is moving from a growth stock model driven by increasing the consumer base to a recurring dividend model driven by services.

This may be easier said than done - here's a deep read on the problems they've encountered trying to get deals with the TV business. Also, it might not be the moment to turn away from the flagship smartphone battle, as it turns out shipments are growing again, and especially Samsung's. Their revenues for Q2 were up 5%, and profits were up 18% - unlike Apple, this implies their margins and average selling prices are improving. Galaxy S7 Edge devices accounted for half the sales by value.

Here's a review of the own-branded devices Reliance Jio is selling for its 4G network. Pricing is as low as $45, reflecting a 4G bill-of-materials premium that may be as low as $5 and significant economies on the network side. Here's a review of the Meizu Ubuntu Touch tablet - it's positive on the hardware, much less so on the OS.

Huawei H1 results are in, and they're decent - revenue was up 40% year-on-year, much of which was down to the consumer business. Huawei is No.1 in China, and it's gaining share off Apple. However, as usual, margins disappointed - the targeted 18% operating margin was missed and they ended up at 12%.

Vestberg goes west; vendors' 5G demo plans; FCC signs off 11GHz of 5G spectrum; Juniper Q2

Huawei isn't the only network vendor with margin problems. Three Ericsson stories follow: Q2 results showed a 22% drop in operating income, driven by a 7% drop in sales and the loss of 9 basis points of margin. Although margins were still 32% - way higher than Huawei's - the balance between relatively commoditised hardware and margin-rich services business is shifting the wrong way. The Swedish newspaper Svenska Dagbladet alleged that Ericsson has been stuffing the channel, recognising revenue even before it billed the client. (It certainly wouldn't be the first time a vendor got in trouble that way - remember Nortel?)  And the CEO, Hans Vestberg, has resigned.

Over at Nokia, meanwhile, everyone seems to be full of beans and dreaming of (what else) self-driving cars. It's about 5G, of course.

In more concrete news, Inside 5G has details of the vendors' plans to demonstrate 5G technology in the US, based on the Special Temporary Authorisation forms they've filed with the FCC. Nokia will be testing something in the 2.1 and 2.6GHz bands at its Sunnyvale site and at Bell Labs through August and September. Ericsson will be showing a 15GHz device for two days next week at T-Mobile HQ. CableLabs is working in the 71-75GHz bands, on what sounds like channel modelling. And National Instruments wants to demonstrate a variety of 5G technology options with its SDR products on the 3.5, 19, 27, 60, and 76-79GHz bands. Nokia also tested a 73GHz system with Bell Canada.

That's only for starters, though. The FCC signed off the Report and Order on 5G spectrum, which finds an additional 11GHz for mobile users like so:

  • The proposal for licensed use in the 28 GHz, 37 GHz and 39 GHz bands makes available 3.85 GHz of licensed, flexible use spectrum. It provides 200MHz block sizes, license areas (Partial Economic Areas), technical rules, and operability across the exclusively licensed portion of the 37 GHz band and the 39 GHz band to make 2.4 GHz of spectrum available. There are  two 425 MHz blocks for the 28 GHz band.
  • Unlicensed use in the 64-71 GHz band makes available 7 GHz of unlicensed spectrum which, when combined with the existing high-band unlicensed spectrum (57-64 GHz), doubles the amount of high-band unlicensed spectrum to 14 GHz of contiguous unlicensed spectrum (57-71 GHz). That 14 GHz band will be 15 times as much as all unlicensed Wi-Fi spectrum in lower bands.
  • Shared access in the 37-37.6 GHz band: Makes available 600 MHz of spectrum for dynamic shared access between different commercial users, and commercial and federal users.

The huge 7GHz of unlicensed up in the 60s is exactly where Facebook's Terragraph mesh-WiGig project is aiming.

Dave Burstein is concerned that the millimetre wave spectrum will require such high network density that it will be a natural monopoly. He also reckons low-band 5G will be coming much earlier everywhere except the US, where the 28GHz fixed-wireless train is puffing down the track. One answer to the competition problem is network-sharing, and here are some details of the new China Tower Corporation.

OFCOM notes that WRC-15 picked several bands for 5G that are used in the UK for fixed wireless, and issues a consultation about its future spectrum needs.

Broadcom reports to the FCC on how the talks between LTE-U proponents and the WiFi industry are going. Two take-home messages: they're going well, but a current sticking point is the definition of "quiet".

Here's an interesting blog post about using Jim Gettys and Dave Taht's Bufferbloat work to fix other performance problems in WiFi.

Juniper's Q2 was OK, although interestingly, it's increasingly reliant on telcos.

Is it 5G use cases you're after? This 95 page 3GPP Technical Report is full of them, says the 3G, 4G, and 5G Wireless Blog.

And is Pokemon Go driving material costs or revenues for the mobile operators? Says Dan Rayburn: no. Presumably that's why T-Mobile USA was able to jump in on the publicity by zero-rating the game.

AWS, Microsoft, Google, Oracle Q2s; CORD gets Linux, Google backing; Amdocs spreads ECOMP

Amazon Web Services saw $2.9bn in revenue for Q2, and $5.4bn for the first six months of 2016. This puts them ahead of their goal of $10bn for the full year, and they intend to open nine new availability zones, in four regions, in the next 12 months.

Microsoft's Q2 results delivered 5% revenue growth. The key contributors were the Intelligent Cloud unit, up 7% at $6.7bn, and Productivity and Business Processes, up 5% at $7bn. (It's worth remembering that Office 365 and Dynamics Online live in Productivity, rather than Cloud.) Meanwhile, "More Personal Computing", which includes hardware and Windows, lost ground. The most purely "cloud" element of Microsoft, Azure, apparently doubled its revenue in Q2, and they are claiming annualised revenue of $12bn for "Commercial Cloud".

Canalys reckons the total cloud infrastructure market was $9.5bn for the quarter, growing 50% annually, with the top four (AWS/MSFT/Google/IBM Softlayer) taking 60.5% of that. If both Microsoft and Canalys' numbers are true, Microsoft must be gaining share rapidly. Looking ahead, they're pushing PaaS rather than IaaS.

Google's Q2s are in. Earnings were better than expected, thanks to both strong sales and lower traffic-acquisition payouts. The train-set businesses in Other Bets are still a money pit, as expected. But the really interesting thing may be that Google's (as opposed to Alphabet's) "Other" line item, which includes their cloud products, has kicked up from 24 to 33% growth, edging over the $2bn/quarter mark, or 11 per cent of their revenue. CFO Ruth Porat says this is down to better sales of enterprise cloud, and gives the credit to their ex-VMWare enterprise chief, Diane Greene.

She also acknowledged the acquisition of Webpass, a small San Francisco-based WISP, and suggested that Google Fiber might use more fixed-wireless in the future as 5G and WiGig become available.

Oracle's results tell a very clear story. Overall revenue was flat for the quarter, down 3% for the year, while cloud was up 49% for both. So they're buying Netsuite, a cloud-based accounting company that Larry Ellison funded back when it started up in 1998, for some $9.3bn.

Intel has open-sourced its Rack Scale Design technology for designing and automating big data centres. That makes two open data centre forums - RSD is parallel to Facebook's Open Compute Project, if it's not on a collision course with it. The good thing about standards is that there are so many of them, as someone said.

Data centre space is selling fast.

AT&T's Central Office Re-Architected as a Data Centre project, CORD, is now an open-source project in its own right, under the aegis of the Linux Foundation. And it's getting support from Google. The next priority is to build a hardware platform for its mobile version, M-CORD, for between $10-15,000. At the same time, the Broadband Forum has joined the project to link in the FTTH and cable worlds. AT&T has also signed up Amdocs as a systems integration partner in order to push its related open-source NFV platform, ECOMP, to other telcos.

On the other hand, Verizon CFO Fran Shammo says the company will make a decision in Q3 about whether or not to sell its 48 data centres.

Verizon, AT&T, T-Mobile, Sprint Q2s; Comcast Mobile

Verizon Q2s are interesting, to say the least. Revenue was down 4% year-on-year in wireless, and group-wide revenues, including fixed and enterprise, were off 5%. On the other hand, postpaid retail net-adds were 615k, implying connections growth of 3.9% year-on-year. That's strong, but it's nothing like the 1.1m achieved in Q2 2015. Wireless EBITDA margin is 47.5%, rising again after a nasty dip last year. But the 3.8% growth in EBITDA margin is not quite enough to keep up with the drop in revenue.

And wireline is net-losing subscribers, even FiOS FTTH subscribers (down 13k, with 41k cancelling video), while its revenue is flat to falling. When you add in the troubled enterprise and wholesale operations, total wireline revenue is off 2.4% year on year, and its EBITDA margin has dived five percentage points this year. So, will there be more FiOS builds like the one in Boston? Lowell McAdam seems to say so, but leaves himself an out.

Against this background, they've decided to spend $5bn buying Yahoo!'s core business.

Dave Burstein quotes Brett Feldman of Goldman Sachs pointing out that there wasn't enough free cash flow in the quarter to cover the dividend payout, and notes that VZ's CAPEX/revenue ratio is falling steadily, which erodes their premium-carrier status. He is deeply sceptical of their ability to run Yahoo! better than its management or sell ads better than Facebook or Google, and points out that there is a long, sad history of telcos losing money in content. VZ's competitors seem to agree.

Speaking of competitors, here's Sprint's Q2. They reported 173k postpaid net-adds, and they claim to have a positive porting ratio for the first time in five years. On the other hand, they net-lost 331k prepaid customers, and the final score of 377k total net-adds includes no fewer than 528k wholesale/M2M. And the company lost $302m in the quarter, compared to $21m a year ago. As predicted, CAPEX is almost at a standstill.

Interestingly, the network management contract with Ericsson is up for renewal, and Sprint is taking part of it back in-house. The original contract was signed in 2009; somehow we doubt, given Sprint's vicissitudes in the meantime, that Ericsson will put it in their sales pitches.

Over at T-Mobile, Q2s are in, and it was their 13th successive quarter with a million-plus net-adds. Specifically, there were 646k branded, postpaid phone net adds plus 476k prepaid, and service revenue was up 12% at $6.9bn, with $225m net income. John Legere said some stuff.

AT&T is planning to speed up its fibre rollout, aiming for 3m passes a year for the next four years, which would be equivalent to 20-25% of their footprint. Except it's not - they're likely to build quite a bit of that out-of-footprint, overbuilding into other operators' markets, notably Windstream and Fairpoint. The plan is fairly simple - the target areas are places where fibre can be deployed for cheap on aerial plant, and AT&T's LTE network already has backhaul.

AT&T's Q2s, meanwhile, show the continuing re-orientation of their Business Solutions division. Overall, its EBITDA and revenue are flat, but underneath that, the cloud-based Strategic Services line is growing fast at the expense of the legacy products. SBS is now 36% of business wireline, or 16% of the total line-of-business, as 55% of it is now wireless.

Fixed broadband revenues were up 15%, driven by the fibre rollout. Although they're keen to hype video, in fact it was up only 0.8%, while the Entertainment line's EBITDA was flat at 25%. Wireless had a 1.4m net-adds quarter, although only 185k of those were branded smartphones. Wireless revenue was down 0.4%, but margins increased 130 basis points.

Comcast Business is still growing like hell. Meanwhile, the parent company has launched a new mobile division and confirmed that it's invoked the option to have an MVNO on Verizon Wireless' network it's had since 2011.

They'll have to contend with Layer3, a start-up cableco (!) that turns out to be financed by Altice.

Facebook, Twitter Q2s; Salesforce dumps all but Samsung and Nexus; the end of Skype P2P

Facebook's Q2s are in, and revenue is up 59% at $6.4bn, while daily-active users are up 15%. To put it another way, their ARPU is rising strongly. Out of that, $2.7bn stayed with the company as operating income, or an operating margin of 43%. 84% of that money comes from mobile ads, and it's another success for their strategy of restricting ad volume and concentrating on quality:

In Q2, the average price per ad increased 9% while total ad impressions increased 49%. The reported increase in price was again driven by the continued mix shift towards mobile where we only show higher price News Feed ads

Twitter, meanwhile, is stagnating, with no user growth to speak of, and still struggling to monetise.

Salesforce is giving up on trying to keep on top of all the Android fragmentation, and will only support Samsung and Google Nexus devices from now on.

Microsoft has announced the end of the native Skype clients for Mac OS X and Linux machines. It's a historic moment - the Windows app was long since rewritten to be a SIP client plugged into Microsoft's client-server infrastructure, and the replacement uses WebRTC, so this means the end of the Skype P2P cloud. Let that sink in for a moment.

Meanwhile, the free upgrade period for Windows 10 is over, and the first anniversary update is coming. Microsoft Stream is their new platform for business video. And Redmond is cutting another 2,850 jobs from the mobile devices business.

ICANN is tearing itself apart over the question of who has the right to the top-level domain .gay - the Economist Intelligence Unit is somehow weirdly involved.

And here's how Pokemon Go was pre-invented in Manchester in 2009.

CMCC adds more fibre than everyone else; new Vodafone AMAP boss; Orange 4G for Africa

Chinese operators are adding fibre at a ferocious rate. China Mobile added 3m net FTTH subscribers in June alone, more than the rest of the world's operators managed together. The previous record was 2.65m for March, and that was China Mobile too. The competition isn't going quite so fast, but they did manage another million between them. Also interesting: UnionPay, the Chinese interbank payments operator, is suffering from competition from mobile money.

Here's a positive take on Reliance Jio. We note there still isn't, still isn't a launch date.

Vodafone has picked former Orange deputy CEO, Vivek Badrinath, to run the Africa, Middle East, and Asia-Pacific division. He replaces Serpil Timuray, who is promoted to group Chief Commercial & Strategy Officer. Both jobs report directly to Vittorio Colao.

Telkom Indonesia reported net income up 33% in H1. Basically everything was good.

Orange's CFO Ramon Fernandez says the firm will have 4G in all its African markets by 2018, and it's looking at investing in FTTH there.

In South Africa, 4G deployment has triggered a surge in data traffic and some of Telkom's 2.3GHz sites are congested.

Oi owns part of Angolan operator Unitel, and it has been trying to get a substantial dividend it claims to be owed paid out. The Angolan shareholders say they offered $900m to buy Oi out, but they refused. Oi says it wasn't them, but something called "Samba Luxco". We can see this being a long one.

And here's a decent Q2 from America Movil.

BT, O2, Orange, Vodafone UK, TI Q2s; Openreach decision

BT's Q2 was, of course, the first with EE consolidated in. It looks like EE had a poorish quarter, as did most of BT except for Consumer, which saw 9% revenue growth as it sold even more football. Interestingly, BT's reporting doesn't give previous years for EE - you wonder why. More importantly, OFCOM has made its mind up, and the big decision is: Openreach stays with BT. That said, Openreach must become a legally distinct subsidiary company, with independent non-exec directors and its own P&L, and BT has been set some investment benchmarks.

This kicks off a consultation period running into October. With a view to that, Vodafone, Sky, and TalkTalk have mounted a new campaign to detach Openreach, called "Fix Britain's Internet". They probably didn't mean this, though. Those three, of course, are the co-investors in CityFibre's FTTH build in York. Light Reading reports that Sky has achieved a 12% take rate in the first three months and is confident of getting to its target of 30-40%.

Orange announced a Q2 much improved by banking the money from the sale of EE. It was the fourth successive quarter of growth at Orange, but the measure of that was probably that the exciting high growth African division grew 2.3%.

Meanwhile, CK Hutchison is suing the European Commission over its decision to block 3UK-O2, while O2 UK's Q2 sales were down 4% in constant currency. Vodafone UK's Q2 was also pretty rough - off 3.2%. Vodafone blames this on the billing system, but it can't be their billing that spoiled O2's day, can it?

Telecom Italia announced what it calls its best quarter since 2009, and signed a joint venture with Fastweb to build more FTTH outside Milan and their existing footprints.

Ziggo-Vodafone.nl looks like it will go through. The Euregulators have cleared DTAG's vectoring upgrade, on condition they accept more duct access, and neither Vodafone nor ECTA are very happy.

The job guarantees offered by Altice when it bought SFR are about to expire, and Altice is now going to cut 5,000 jobs. The French government is not happy either.

And you just knew Free's phasing out of national 3G roaming would be quirky. Free intends to cap the speeds offered to roaming users at progressively lower levels. Presumably the idea is to encourage holdouts to get 4G?