A hard quarter in the West; Microsoft's reboot?; Amazon's new EC2 - Telco 2.0 News Review
- Smartphones: Nokia Q2: bad, but not *that* bad
- Broadband Connectivity: Q2s: Ericsson, ALU, Vodafone, no fun
- Voice 2.0: Minutes falling in the UK, exit WAC, re-enter Enyo and Phono
- Google: Exit Marissa Mayer, CAPEX numbers
- Microsoft: takes loss with write off, but is it attempting to reboot with Windows 8?
- Cloud Special: Amazon's latest - less cloudy, very very fast, Apple goes for smaller DCs?
Horace digs into the Nokia Q2 figures and reckons that the Lumia phones aren't doing so badly in the US, specifically in terms of dragging the average selling prices upwards. However, in terms of volume, it's nothing to set the world on fire, and he concludes that the biggest problem for Nokia is time. Of course, part of the problem there is that Microsoft's Windows Phone 8 won't run on the Lumias, which leaves 4 million orphan smartphones in the market.
The headline numbers are here and are pretty grim, with a substantial net loss and sales sliding. Asia-Pacific is the only growth market. The stock market was surprisingly favourable, presumably on the basis that it could be worse. Interestingly, the burn-rate has dropped sharply in the last three months, which suggests that the fight for time is getting somewhere.
Elsewhere in smartphones, the key to Apple's market share turns out to be nothing else than good old carrier subsidy, with every $100 bung translating into 4-5% of market share in the United States, and a cushion of about 6.3% Apple cultists at a zero subsidy level. Hence, the Chart of the Week:
Huawei's Ascend P1 climbs down off that big horse from MWC and lands in the UK, along with ZTE's Grand X. HTC reduces its investment in Beats Audio. Samsung gets a win in court, as a judge orders Apple to state publicly that Samsung didn't copy the iPad. They also buy most of CSR.
Sensible idea of the week: Qantas chucks out the sporky in-flight entertainment terminals and gets iPads instead. If you're in business class.
It was a heavy results week. Ericsson Q2s were out, and profits were down 63% on poor sales in the networks division and a loss at the STMicro joint venture. Alcatel-Lucent announced a small loss for Q2 and warned on its profits for the year. Of course, last week saw horrible numbers at ZTE, so the whole infrastructure sector is looking sad.
Vodafone Q1s were out, and the pattern from last time was clear - the Southern European markets are getting a battering from the macroeconomic crisis. While service revenue in Germany was up 4.2%, it was down 7.7% in Spain and 10% in Italy, with the UK fitting in between at a 0.8% drop. The emerging markets, notably Turkey, India, and South Africa, did well and overall the company just managed to inch ahead by 0.6% on an organic basis.
The Australian division is still struggling but the Voice 2.0 project OneNet is now up to 2.2 million users and is supporting strong growth in the enterprise segment. Data revenues were up everywhere, but the big question there is margins.
Tele2's story is pretty clear - sales were up 10% but EBITDA down 3%, due to the spending on marketing that got the additional customers through the door.
Verizon Wireless turned on more LTE, and managed to up its ARPU by pushing more data. Although the caveat about margins is still relevant, CAPEX was down 2.1% after last year's LTE build-out spree. Dave Burstein reports that Verizon is going to get rid of its copper where the FiOS fibre runs and perhaps also where the LTE coverage reaches.
Brazilian regulators stop the mobile operators selling at all in much of Brazil. Ouch. BT gets a contract for Welsh fibre.
OFCOM released its annual data-dump this week, showing voice usage falling in the UK. The drop is faster on fixed than mobile, and as a result, there were more mobile minutes than fixed for the first time ever in 2011. 44% of adults have used an OTT messaging option, and 22% VoIP, but despite that SMS traffic is still growing.
Here's an interview with Rebtel's CEO Andreas Bernstrom, who argues that VoIP at least ought to be about call quality rather than just free on-net calls and praises Telefonica Digital's TuMe.
Is AT&T going to start charging extra for Apple FaceTime calls?
The good people at Phono have updated their softphone SDK to support Adobe's voice streaming protocol, which seems to be the first in their stack to include acoustic echo cancellation.
It looks like WAC has come to the end of the road, with Apigee buying the network-side technology and the rest being "folded into the GSMA".
On the other hand, the Enyo HTML5 framework, perhaps the most interesting bit of WebOS, lives on. HP announced version 2.0 this week. And here's an interesting new tool for animations and GUIs in HTML5. Both Android and BB OS 10 updated their SDKs this week.
Data Center Knowledge updates its valuable series of Google CAPEX figures, putting last quarter at $774m, which is certainly plenty but nothing like the massive surge at the end of 2010.
Marissa Mayer, most recently the head of local, maps, and location at Google, but better known as the company's user-interface guru when Google was known for really nice user interfaces, has unexpectedly quit and taken on the challenge of being Yahoo!'s fifth CEO in as many years. While also having a baby. No pressure, like.
ZDNet summarises the challenges ahead as relevance, innovation, definition, content, and morale, which anybody would consider a packed inbox. If there's anything Mayer's arrival will help with in itself, Wired argues, it's convincing the best talent in Silicon Valley that there's any point joining Yahoo.
Of course, as they point out, even in the last few years Yahoo! engineering has had some major achievements - the now-ubiquitous Hadoop data-mining systems started out there, Pipes and YQL are impressive rapid development tools, and the YUI user-interface framework is considered a classic. And they invented the hackday.
Mayer is mostly remembered for the ultra-minimal look and feel of classic Google products, and the exhaustive A/B testing process that underlay them, but her original specialisation was artificial intelligence. Gawker seems to have overstated that a bit: Hologram or Android?
Meanwhile, Wired visits a Facebook hackathon and gets all excited, while the New York Times reports on scepticism about how useful Facebook ads really are. It's a nice demonstration of how the company sits between Valley/venture capital optimism and Madison Avenue/Wall Street cynicism. We'll all find out on Thursday, when they publish earnings numbers.
If the Facebook IPO didn't give you enough of a speculative kick, two biggish tech IPOs are coming up - the (rather good) travel meta-search engine Kayak, and software-defined network vendor Palo Alto Networks.
Microsoft, for its part, announced its first quarterly loss ever this week as it wrote down the value of the Aquantive advertising business it bought in 2007 for a mere $6bn.
Search Engine Land has a fascinating analysis of Microsoft's search alliance with Yahoo!, arguing that it's a major disappointment. Yahoo! outsourced search to Microsoft Online Services some years ago, basically becoming the launch-customer for Bing. Under the terms of their agreement, MS has to deliver revenue per search similar to Google's, or pay the difference in cash. A large gap has opened up, which MS has to pay. MS has frequently tried to get out of this, but Yahoo! has held them to the deal. SEL thinks that a break point is approaching, where Yahoo! will be sufficiently desperate to walk from the deal at the same time as MS gets sick of forking over the guaranteed payments.
The question is, then, whether Yahoo! licenses Google Search, whether it buys a startup (Blekko, DuckDuckGo?), or sets about rebuilding its search product. Marissa Mayer's arrival adds an unknown - as a Googler, would she be open to a Google-Yahoo! deal? Or would she relish the massive scaling exercise of bringing a search startup to the Yahoo! front page? Or even the epic challenge of a completely new search product?
But that wasn't the biggest news at Microsoft. Jeff Atwood argues that Windows 8 is the first genuinely innovative product from them since Windows 95, a bet that will either secure their future or else plunge the company into deep crisis. And if it works, it may well be because it takes the hardware market with it away from classic PCs. ZDNet argues that MS is focusing on change in the consumer market above all.
A lot of things have changed at Redmond. For one, the MS Azure cloud developers are substantial contributors of code to the Linux kernel these days, which would have been heresy a few years ago. Even if sometimes, as the Wired enterprise blog reports, their sense of humour is a bit puerile. Azure is very big, although Data Center Knowledge points out that MS's numbers include all their internal users and may double-count objects that appear in different application layers.
MS aren't taking any chances in re-engineering Windows for tablets, mobile phones, and the Web. For example, check out this forensically detailed blog post about designing the touch keyboard.
On the other hand, 20% of the world's supply of stolen logins match a Microsoft account, and interviews with Microsoft staff suggest that Steve Ballmer's insistence on a curve-grading rank-and-yank assessment process in every workgroup is responsible for Windows 7 looking so much like Win95.
Here's a bit of Google customer service: don't bother phoning, but if your blog post goes viral...
The big news in the cloud this week is Amazon Web Services' new EC2 option, which lets you configure EC2 servers with Flash-based solid-state disks in case you need really fast I/O. This is obviously an interesting use of the cloud - isn't the point that you don't need to think about hardware? - and one of the places where the boundary between cloud and dedicated hosting is hard to determine.
James Hamilton has more technical detail. Netflix's Adrian Cockcroft has a long, detailed, and interesting post on performance testing the new instances. He points out that they made it possible to run one of their most demanding applications with substantially less complexity, as it no longer needed an in-memory cache.
However, the secret sauce here is that the access to the disk is purely local, rather than being somewhere in the cloud over the network. In a sense, there's a trade-off between the "cloud" element and the "blindingly fast" element. Hamilton points out that one of the main use cases is moving databases into the cloud that use a sharding architecture, splitting the data into independent subsets. If the data is stored locally, you can't expect the cloud to look after it...so the sharding and replication is vital, in case of a node failure. But wasn't part of the reason we moved into the cloud to avoid the complexity that involves?
Over the wall in the OpenStack world, the TryStack prototype cloud is running on ARM low-power chips, and Cisco is putting more SSDs into servers. Intel's Q2 results show 14% growth in the data centre group.
And this is interesting: Apple has filed the planning application for the next phase of its iDatacenter. But rather than another 500,000 square foot behemoth, they're looking at a 21,000 square foot "tactical modular data centre", possibly using prefabricated pods. Have data centres got to the point where, like oil tankers, the economies of scale run into practical limits?
Marks & Spencer is preparing to spend £100m moving their IT systems out of Amazon and back in-house.
Also, did you know that the current iteration of OpenStack is codenamed "Folsom", like the prison? If you did, you might be less surprised by the latest startup incubator project: the one aimed at the inmates of San Quentin.
O2 makes nice, offering the outaged a £10 service credit. Vodafone femtofail. Syria briefly vanishes from the Internet, all but five networks - and one of those hosts a fake Skype encryption tool. The Internet traffic jam. All the supposed tech laws in one place.