Amazon & Baidu Q2 17A – Back to basics

Reply to this post






Amazon and Baidu get back to what they do best.

Amazon – Not making money

  • Amazon reported disappointing results and guided weakly as it once again spent everything it could on investing in future revenue growth.
  • Q2 17A revenues / EBIT were $38.0bn / $628m compared to consensus at $37.2bn / $1.0bn.
  • AWS put in another mighty performance with revenues of $4.1bn and margins of 22% but this was gobbled up by the international operation where margins have fallen to -6.3% from -1.4% in Q2 16A.
  • I am pretty sure that is mostly driven by Amazon’s absolute determination not to lose India to Flipkart / Snapdeal the way it lost China to Alibaba.
  • The good news is that Flipkart and Snapdeal are squabbling over their merger and the longer it takes them to get it done, the less chance they have to keep Amazon out of their home market. (see here).
  • This heavy investment looks set to continue with Q3 17E guidance disappointing once again.
  • Q3 17E revenues / EBIT are expected to be $39.25bn – $41.75bn ($40.5bn) / LOSS $400m – $300m (LOSS $50m) compared to consensus at $39.9bn / $1.1bn.
  • There is no sign of this “bumbling around break-even” in sight and consequently the valuation of Amazon looks more stretched than ever.
  • I prefer not to pay now for profitability that very fleetingly materialises.

Baidu – Performing in line with China.

  • Baidu reported good Q2 17A revenues as the regulatory impact on its revenues has past and the company kept a tight lid on expenses.
  • Q2 17A revenues / net income were RMB20.7bn / RMB4.4bn compared to consensus at RMB20.7bn / RMB3.3bn.
  • A large part of this improvement has come from cutting back on investments in its food delivery business but also from a big fall in traffic acquisition cost which fell from 15.9% of sales in Q2 16A to 11.9% in Q2 17A.
  • Despite the cuts, investments in AI and content remain intact and are the two main thrusts for revenue growth beyond search.
  • In AI, I rank Baidu highly, although it is very focused on China, and think that this is its biggest strategic advantage to remain a big player in Chinese Internet.
  • Baidu should be able to make its services more intuitive and useful compared to those of its competitors which should help its services win and keep more users.
  • The outlook for Baidu remains steady, now that the regulatory problem is in the rear-view mirror, and I see an upwards correction as it catches up with its peers.
  • I continue to like Baidu alongside Microsoft and Tencent.


Facebook Q2 17A – Cash community

Reply to this post






Community is about cash generation.

  • Facebook reported another set of excellent results and laid out pretty concrete plans of how it aims to put off, for as long as possible, the inevitable slowdown in its growth.
  • Q2 17A revenues / adj-EPS were $9.3bn / $1.32 compared to forecasts of $9.1bn / $1.32.
  • Mobile advertising, and especially video, underpinned most of the growth as mobile now accounts for 87% of total advertising revenues.
  • Facebook now has 2.01bn MaU of which 1.32bn visit everyday
  • Two major themes have emerged over the last 6 months which were further emphasised at these results.
    • First: community. Facebook is moving away from connecting friends (as it has already done this) and towards creating communities.
    • There are already 100m members of groups around particular interests which Facebook aims to push much higher.
    • These groups meet both virtually and physically and I think they represent an incremental monetisation opportunity.
    • This is because they represent the most engaged users where their interests are as clearly defined as they are on Twitter.
    • This means that Facebook should be able to monetise them much more effectively as the advertisements served will be more relevant and therefore can be more highly priced.
    • Furthermore, should Facebook succeed in growing the membership of these groups meaningfully, the average time spent by users on Facebook will also rise.
    • This will give both a price and volume lift to revenues allowing much faster growth.
    • Second: Artificial Intelligence. It is clear that this is Facebook main strategic priority.
    • This is because it is very far behind the curve when it comes to the quality of its AI, and it badly needs to at least be able to understand and categorise the huge amounts of data that it generates every day.
    • Facebook also intends to use AI to understand its users better so that it can suggest content that exists outside of their social circle in which they might be interested.
    • This will also have the convenient side effect of enabling Facebook to target its users more effectively, thereby increasing the price it can charge to marketers.
    • AI still remains a major weakness for Facebook but importantly, it is aware of the problem and is working on fixing it as fast as it can.
  • Behind the desire to connect people and create communities lies a Sheryl Sandberg’s highly efficient cash collection machine.
  • At the end of the day Facebook is a business not a philanthropic organisation and it is doing an excellent job of earning a good return from the data it collects.
  • I still remain concerned with short-term growth due to the maturation of its core businesses and the fact that new areas like messaging and gaming have yet to really generate revenues.
  • So far in 2017, Facebook has been able to defy both its own (and my) forecasts for revenue growth but the comparisons are getting tougher and tougher to keep up this pace.
  • Hence, I doubt that Facebook can keep this up until its new businesses mature and I still see a pause coming in revenue growth over the next 6 to 9 months.
  • It is at that point, that I am looking to get in for the long term as Facebook is showing all the signs of becoming the biggest ecosystem of them all.

Facebook – Empty head

Reply to this post






Another smart speaker that badly needs a brain.

  • It looks like Facebook is joining the ever more crowded smart speaker bandwagon, but without a decent brain inside the box, it may as well be a paperweight.
  • One possibility is for the device to use Cortana as it comes from one of the few companies that doe not compete directly with Facebook: Microsoft.
  • The device looks like it will be using a 15-inch screen from LG and will be manufactured by Pegatron but beyond that there are very few details.
  • I suspect that Facebook may be trying to take a slightly different tack here.
  • This is because:
    • First: the smart speaker market is already very crowded,
    • Second: Facebook has no brain of its own to install in the box,
    • Third: Facebook is more focused on community than smart home.
  • Facebook’s main objective in life is to bring its users closer together using its apps and to give them a sense of community.
  • While this all sounds great for users, the reality is that they will end up spending more time inside Facebook’s fledging ecosystem, generating more traffic and thereby increasing Facebook’s ability to make money from them.
  • Hence, I suspect that this device may be aimed more at making it easier for Facebook friends to spend time with each other by voice, video, messages or even images.
  • However, to earn a place on the increasingly crowded countertop of consumers, it is going to need voice functionality of some description.
  • I think that Facebook M, which is Facebook’s own digital assistant is hopelessly inadequate to fulfil this role, meaning that Facebook will have to get one from somewhere else.
  • Top of my list for this is Cortana which, while not the sharpest tool in the box, it is the only one whose owner is not competing directly with Facebook.
  • In fact, I have seen Microsoft and Facebook creeping closer together (see here) over the last few years and this is a collaboration that could make some sense.
  • With a bit of tinkering on Microsoft’s part, Cortana could be taught how to deal with the majority of the tasks that users ask smart speakers to perform.
  • This work is probably already going as Microsoft may already be working on a smart speaker of its own.
  • Combining this with the screen and Bing would give the device a reasonable shot at doing a decent job of answering queries.
  • This is just another example of how badly Facebook needs to bring its AI up to a level at which it can compete on a level playing field with Google.
  • This would also help Facebook deal with the objectionable content problem that it has on its platform as its current answer to this is to throw more humans at the problem.
  • For me, this has to be Facebook’s number one strategic priority and the progress displayed at F8 on image and video recognition was somewhat encouraging (see here).
  • I am still quite cautious with regards to Facebook’s outlook for this year as I don’t think that either its video offering or its gaming offering are mature enough to bring the company back to high growth in 2017.
  • This combined with requirement to really improve its AI to compete with the other digital ecosystems leads me to still prefer Baidu, Tencent and Microsoft.

Alphabet – Other people’s money

Reply to this post






Core businesses showing signs of maturity.

  • Alphabet reported good Q2 17A results but the fact that it has having to share more of its revenues with third parties is hurting profitability and indicates that growth in its core properties is maturing.
  • Q2 17A revenues-ex TAC / Adj-EPS were $20.9bn / $8.90 below consensus of $21.2bn / $10.34.
  • While I consider the EU fine to be little more than an annoyance (see here), it is the traffic acquisition cost (TAC) where I have concerns.
  • Over the last 12 months, Google has been forced to pay away more of its revenues to its network members and distribution partners.
  • Google network members are now keeping 72% of their turnover compared to 70% a year ago and distribution partners now get 11% of the revenues that Google generates from their devices compared to 9% a year ago.
  • The problem here is that this falls straight to the bottom line and looks to have been entirely responsible for the weakening of margins experienced during the quarter.
  • To a certain degree, what Alphabet is doing is buying growth in revenues and paying for it with lower profitability.
  • This is exactly what Yahoo did much more aggressively prior to its acquisition by Verizon and it was able to mask a decline in its core business by buying in revenues from elsewhere.
  • It implies that monetisation of its own properties and content such as search, mail, maps is starting to hit their capacity ceilings, leaving the company needing to find more growth from third parties.
  • In this regard, the star of the quarter was YouTube which now has 1.5bn MaU with a staggering average engagement time of 60 minutes per day.
  • YouTube is the main home of the alternative media which continues to gain substantial traction across the world as well as the place to go to learn how to do almost anything.
  • YouTube has also seen very high growth of viewing from regular TV screens and is gradually making the move into traditional entertainment.
  • Machine learning and AI are found everywhere in Google’s products and given its lead here, it has the ability to make its services more useful and more intuitive than the competing services of other ecosystems.
  • Hence, I think the outlook remains pretty good for Alphabet but the underlying growth in the core business looks like it is maturing.
  • Hence, profit growth could lag revenue growth as margins come under pressure from increasing TAC leading to more disappointment.
  • I remain pretty cautious on Alphabet overall as the share price has more than kept up with the growth the company has enjoyed over the last 24 months.
  • I still prefer Tencent, Baidu and Microsoft.

Baidu – Talking machines.

Reply to this post






The way in China is wide open for Baidu.

  • Baidu’s strategy around its AI platform and its Duer OS has become clearer and with the support of a large number of chip vendors, it is in pole position to be a major player in smart connected devices in China.
  • DuerOS is not a traditional OS like Android or iOS but instead is a much more focused sub-system that is capable of bringing intelligent voice control and intelligence to any device in which it is implemented.
  • DuerOS’s direct comparisons are the software that runs on Amazon’s family of Echo products, Google Home or’s DingDong.
  • While Duer still speaks no English at all, I think it is currently by far the leading contender in this category for China for two reasons:
    • First: ecosystem. Baidu has already lined up an impressive list of component and device manufacturers who will be implementing DuerOS in their products.
    • Realtek, Intel, Nvidia, MediaTek, RDA, Conexant and ARM have signed up to support the system, which combined with a series of device makers, should create a pretty healthy ecosystem.
    • There are already around 30 products in the pipeline encompassing pretty much the entire range of domestic electronic devices and appliances.
    • Second: Artificial Intelligence. RFM research (see here) has indicated that Baidu’s AI is second only Google and certainly far better than anything else currently on offer in China.
    • This is a product of years of work as well as having developed by far the leading search function in China.
    • The net result is that DuerOS, like Google Assistant, should be able to provide users with the best experience when it comes to understanding and dealing with voice based requests.
  • Putting these two together put Baidu in pole position when it comes to creating an ecosystem within which a whole series of devices can talk and understand both the user and each other as well as work together.
  • This represents a big threat for Xiaomi which has laso built quite a large ecosystem of smart devices but they really lack the intelligence that DuerOS can offer.
  • The upside for Baidu is that by powering all of these voice-enabled gadgets, it will be able to gather data about its users that it will be able to make its search all the more relevant.
  • One of the big differences between China and Western markets is that no one seems to care very much about privacy (see here) meaning that this strategy could work very well.
  • I don’t expect Baidu powered machines to suddenly start spewing out voice-based advertising but learning what its users like and what their needs are will help it make its search results more accurate and hence more valuable to advertisers.
  • Baidu is still the search leader in China but its recent problems with fake advertising are only just behind it and this could provide a good pillar for long term growth.
  • I think that its real rivals, Alibaba, Xiaomi and Tencent, are miles behind when it comes to AI and voice-based services, leaving the Chinese market wide open for Baidu.
  • This combined with its leadership position in AI and search are the main reasons why I still like Baidu together with Tencent and Microsoft.

Alibaba – Virgin home

Reply to this post






Another day, another home assistant.

  • Alibaba is moving to launch a home speaker, much like Amazon Echo, that should direct users to its e-commerce services but I suspect that if it teams up with Baidu, it will launch a better product
  • The obvious intention is to make it even easier for users to shop on Alibaba’s sites but given the differences between China and developed markets, I am not convinced that this is necessary.
  • Alibaba’s users are already making over 80% of their purchases on mobile devices which is much higher than Amazon or other e-commerce platforms in developed markets.
  • The reason for this is that China is a mobile first market (see here) because in China, the mobile internet works much better than fixed.
  • In developed markets, the opposite is true which has meant that shopping via a PC or Mac still represents the majority of transactions.
  • Hence, if one can easily order products using voice commands with Amazon Echo, it represents an easier experience than using a PC and a web browser.
  • However, apps on smartphones are so optimised for the task for which they have been designed that it may not end up being much easier to use a speaker from Alibaba than the mobile phone.
  • This will especially be the case if the assistant that Alibaba puts into the speaker is not that smart.
  • RFM research (see here) has not highlighted Alibaba as being a leader in AI meaning that the intelligence of its speaker is likely to be second or even third rate.
  • Hence, if the shopping experience is not much enhanced by having a speaker, it will need to be very good at other functions in order to be appealing.
  • Typically, these have included the ability to play music, answer general inquiries and control smart devices around the home.
  • Offering a decent user experience in these areas is a much more general AI problem and one with which I think Baidu is far more advanced.
  • Hence, I think that if Alibaba comes to some arrangement with Baidu to use its personal assistant Duer for part of the functionality, it will end up with a much better user experience.
  • The good news is that China is almost virgin territory when it comes to this space as I do not believe that either Baidu or have had any real market impact with their products.
  • Furthermore, the failure of Amazon in China and the blocking of Google services has meant that foreign competition is almost non-existent.
  • This means that even if Alibaba’s speaker is sub-par due to the basic level of AI that Alibaba has to put into its offering, it may still sell reasonably well with the right marketing push behind it.
  • I think that Alibaba is also still well behind when it comes to the smart home and so it would make sense to emulate Amazon’s extremely developer friendly attitude.
  • Xiaomi has also built up a reasonable ecosystem of smart home devices In China that use its API, and so including this at the factory would also seem to be a good idea.
  • I have been quite encouraged by Alibaba’s emerging understanding of the importance of data and how it can benefit by integrating it all together and making as much use of it as possible.
  • This move with a home speaker would represent a further expansion of that understanding.
  • I still struggle with the valuation of Alibaba as there is a lot built into its share price, but fundamentally it is doing all the right things to be one of the big ecosystems at home.

Digital sensors – Eyes and ears.

Reply to this post






Sensors are the eyes and ears of AI.

  • Data is the raw material of artificial intelligence (see here) meaning that will be increasingly critical that the sensors that collect that data are reliable and accurate.
  • Nowhere is this more true than in eHealth where inaccurate data is useless at best and deadly at worst.
  • This is why there is still a big market for extremely expensive medical monitoring equipment but I see signs everywhere that this is starting to come to an end.
  • This also explains the problems that the likes of Fitbit, Apple, Xiaomi, Garmin and others are having as the data they generate is such low quality that it can really only be used for recreational fitness.
  • I see two ways in which the data that these sensors generate can be improved.
    • First: Improve the quality of the sensors themselves.
    • If an optical heart rate sensor can gather data as reliably and as accurately as an ECG, then this would have substantial ramifications for cardiac medicine.
    • Not only could the equipment costs be slashed, but high-risk patients could be continuously and un-invasively monitored allowing many cardiac events to be predicted and stopped before they occur.
    • The sensor industry is feverishly working on this with the latest launches promising more and more accuracy and detail.
    • Despite this, I have yet to meet a cardiac sensor company that is claiming that it can hit the kind of quality that would allow it to be certified with the FDA.
    • The same is not true in blood pressure where small start-up Leman Micro Devices is making some bold claims.
    • It has come up with a tiny blood pressure sensor that can fit onto a smartphone which it thinks is very close to being good enough to measure blood pressure at a medical grade with FDA approval.
    • Second: create intelligent software that improves the quality of the data.
    • There are many examples of algorithms being used to meaningful conclusions from low quality data both in and out of the medical field.
    • In automotive, retro-fitted vibrations sensors are being used to track the condition of tyres, wheels, shock absorbers, brakes and the steering wheel (see here and here).
    • This is not because a great sensor has been invented, but because these companies have worked out to interpret the data that most people consider to be random noise.
    • Phillips is also quite good at this which is why its health watch is recognised to be generating good quality data even if it does struggle in other areas.
  • Hence, I see the road to accurate data being made by eroding the problem from both ends combining both better hardware sensors and much better software to interpret the signals.
  • I think that this is crucial as sensors are the eyes and ears of the machines upon which the world increasingly depends.
  • Consequently, I think that sensors will remain an area of intense investment and an area where I would want to be invested.
  • The issue of course is to separate the solutions that have real prospects from those that are merely riding the wave of hype and easy investment.


Facebook – Dial M pt. II

Reply to this post






Very little AI in the assistant that has launched.

  • Now that Facebook’s digital assistant is available in the wild, one can see how simplistic it is indicating just how far Facebook still has to go to get a real grip on making its services more intelligent.
  • Facebook M has been in beta for over 18 months and has comprised of a combination of automated responses and human interactions where the vast majority of the tasks have been carried out by humans.
  • The problem with using humans of Digital Life services is that it is very expensive to scale the service for 2bn users especially when the service will be funded by advertising.
  • This why Facebook is working as quickly as it can to develop its in house expertise and while it remains a laggard in AI, it has shown some progress.
  • For example, at its developer conference (see here), it showed some good progress on machine vision enabling its apps to recognise the world they can see through the smartphone camera.
  • It also made Facebook M available to US users and most recently in Spanish to users in Mexico and US.
  • However, what has gone live is only a small part of the grand plans that were announced in 2015 which had an always on, all knowing bot with which the user could do almost anything.
  • Instead, Facebook M is limited to suggestions referred to as “M suggestions” which are contextually sensitive pop ups that appear in messenger when the user types messages.
  • For example, hello (or hola) results in the suggestion of emoticons that are waving or “tomorrow” can result in the suggestion of a link to the calendar to create an appointment.
  • The available functions are very limited leading to believe that each function has been manually programmed using statistical analysis meaning that there is virtually no AI in the service that has launched.
  • Although, the service is extremely limited at present, Facebook has created a placeholder ready to be upgraded when its ready, as well the possibility to generate some data that should help improve what is already there.
  • Most of the AI that I can see in Facebook is in machine vision where Facebook demonstrated some progress at F8 (see here).
  • However, outside of mixed reality, the immediate applications for this in Facebook’s ecosystem remain quite limited.
  • This reinforces my opinion that Facebook is way behind when it comes to AI and that the biggest challenge it faces is to bring its AI into line with that of its main rivals.
  • The problem is that its rivals are starting to use AI to improve the depth, richness and utility of their services potentially leaving Facebook behind.
  • To keep up, Facebook currently throws humans at its AI related problems (eg fake news and objectionable content) which is clearly not scalable.
  • Unless the AI problem is fixed, Facebook will have to employ more and more humans leaving its EBIT margins, valuation and competitiveness at risk.
  • Facebook has some time to address this problem as its newer Digital Life services of Gaming and Media Consumption have scope to keep revenue growth going in the medium term (see here).
  • This is why I like Facebook’s investment potential but I am waiting for the short-term fall in revenue growth (see here) to be priced in before pulling the trigger.

Alphabet – Goodbye blue sky pt. II.

Reply to this post






Homeless robots find permanent shelter.

  • Alphabet has reached a deal to sell both Boston Dynamics and Schaft to SoftBank leaving it more focused on its core business of collection and monetisation of Internet data.
  • Boston Dynamics is a robotics company that specialises in robots that are autonomous as far as navigating and adjusting to their immediate environment.
  • SoftBank is also acquiring Schaft from Google which is a humanoid robotics company that was spun out of the University of Tokyo.
  • These robots can move around with relative ease but how they would be able generate value for Alphabet shareholders was always unclear.
  • At the end of the day Alphabet is a data and analytics company whose objective is to categorise and understand every piece of digital information that users generate and to sell those insights to marketers.
  • Every other piece of hardware that Alphabet makes from Google Home to Pixel and Internet Balloons, have the capacity to collect huge amounts of data and thereby generate can value to the core business.
  • Autonomous robots that can carry out physical tasks do not generate data about users because they are designed to replace them making them a bad fit inside Alphabet.
  • Furthermore, the robotics effort at Google was the brainchild of Andy Rubin and his departure, combined with the much greater focus on fiscal discipline, meant that the robots became homeless inside Alphabet.
  • I have long believed that Boston Dynamics will be much more at home inside a company that can make use of them.
  • Good examples of this are Amazon and Alibaba for logistics or someone like DHL or UPS.
  • Softbank is a good example of this but also has the benefit of a very long-term mindset when it comes to its strategy.
  • SoftBank already produces the Pepper robot which is supposed to be able to read human emotions and help shoppers when they enter a shop or place of business.
  • I met Pepper when wandering the halls of Mobile World Congress and CES and have to admit I was not that impressed by what it was capable of.
  • Consequently, it looks like SoftBank needs to really beef up its robotics expertise if it wants to be a player in this space which is what these two acquisitions should start to accomplish.
  • Hence it looks like this acquisition will not be part of the $93bn Vision Fund but instead be part of SoftBank itself.
  • Boston Dynamics, Schaft and I suspect SoftBank’s own robotics division have been struggling to find ways to generate revenue necessitating a home with a very long-term view.
  • That home used to be Alphabet, now it is SoftBank.
  • The sale of these two businesses will further boost Alphabet’s short term financial performance but I continue to think that all of the recent fundamental improvement in Alphabet is more than discounted in the share price.
  • Hence, I continue to prefer Tencent, Baidu and Microsoft.

Samsung Bixby – Failure to launch

Reply to this post

RFM AvatarSmall






Bixby is not fit for purpose.

  • Samsung has once again delayed the roll out of the voice component of its digital assistant Bixby further reinforcing my opinion that Samsung can really only compete in hardware.
  • This, combined with the poor performance already offered by Bixby services on the Galaxy s8, leaves me unsurprised that a method to rewire the Bixby hard key to Google Assistant has already been published.
  • Bixby was launched with much fanfare at the unveiling of the Galaxy s8 and promised the following:
    • First, completeness: This promises to give users complete control of enabled apps rather than the few tasks offered by other assistants.
    • Second contextual awareness: Samsung is promising that Bixby will be aware of the context within which it has been triggered, making it more relevant and useful.
    • Third natural language recognition: Bixby should be able to understand complex, multi-part questions as well as prompt the user to clarify the pieces that it does not understand.
  • I have been testing Bixby extensively and so far, the experience bears no resemblance whatsoever to these promises.
  • Instead Bixby offers a series of suggestions of videos to watch and articles to read that bear little relevance to any of my interests or my history.
  • The one thing that Bixby can get right is to highlight which apps I use most but the functionality of suggesting which app I am likely to want to use next based on the time of day or my circumstance is nowhere to be seen.
  • These features are very similar to those promised by Viv, the artificial intelligence company that Samsung purchased in October 2016 which is clearly the source of this product.
  • However, It appears that Bixby as it exists today has nothing to do with Viv which partly explains the poor functionality but also makes me wonder why Samsung acquired it in the first place.
  • This is a sure indicator of just how far behind Samsung is compared to everyone else when it comes to developing intelligent services.
  • RFM research (see here) has identified three stages of voice recognition of which the first and by far the most simple is the accurate conversion of voice to text.
  • Almost everyone, even Facebook, has pretty much cleared this hurdle but it appears that Bixby still has not.
  • Furthermore, Bixby vision is also way behind the curve as it is unable to properly identify objects.
  • Instead what it does is search Pinterest for other pictures with similar pixel patterns.
  • It does not identify objects nor offer any real functionality beyond finding similar pictures rendering it useless.
  • Even Facebook, which I have long identified as being behind in AI, is demonstrating reasonably good machine vision which leads me to put Samsung far behind even Facebook.
  • This leaves Samsung exactly where I left it as a manufacturer of excellent but commoditised hardware that outsells it nearest competitor by more than 2 to 1.
  • As long as it can maintain that edge, I have no fear for its handset margins but Huawei is trying very hard to close the gap.
  • Huawei’s disappointing handset performance in 2016 has led it to be more focused on profitability this year meaning that it will not be trying to turn the screws on Samsung with quite the same vigour.
  • Hence, I think that Samsung is set up to have a good 2017 but the rally in the share price has more than taken this into account.
  • Hence I continue to prefer Microsoft, Tencent and Baidu.