IBM has announced the completion of the acquisition The Weather Company’s B2B, mobile and cloud-based web-properties, weather.com, Weather Underground, The Weather Company brand and WSI, its global business-to-business brand.
At first blush this may not seem like an obvious pairing, but the Weather Company’s products are not just their free apps for your smartphone, they have specialised products for the media industry, the insurance industry, energy and utilities, government, and even retail. All of these verticals would be traditional IBM customers.
Then when you factor in that the Weather Company’s cloud platform takes in over 100 Gbytes per day of information from 2.2 billion weather forecast locations and produces over 300 Gbytes of added products for its customers, it quickly becomes obvious that the Weather Company’s platform is highly optimised for Big Data, and the internet of Things.
This platform will now serve as a backbone for IBM’s Watson IoT.
Watson you will remember, is IBM’s natural language processing and machine learning platform which famously took on and beat two former champions on the quiz show Jeopardy. Since then, IBM have opened up APIs to Watson, to allow developers add cognitive computing features to their apps, and more recently IBM announced Watson IoT Cloud “to extend the power of cognitive computing to the billions of connected devices, sensors and systems that comprise the IoT”.
Given Watson’s relentless moves to cloud and IoT, this acquisition starts to make a lot of sense.
IBM further announced that it will use its network of cloud data centres to expand Weather.com into five new markets including China, India, Brazil, Mexico and Japan, “with the goal of increasing its global user base by hundreds of millions over the next three years”.
With Watson’s deep learning abilities, and all that weather data, one wonders if IBM will be in a position to help scientists researching climate change. At the very least it will help the rest of us be prepared for its consequences.
New developments in AI and deep learning are being announced virtually weekly now by Microsoft, Google and Facebook, amongst others. This is a space which it is safe to say, will completely transform how we interact with computers and data.
Against that backdrop, it is heartening to see some more enlightened cloud companies doing the right thing. Salesforce announced today its second renewable energy purchase agreement. The first announcement, made just last month was of the signing of a 12-year wind energy purchase agreement, for 40MW of a new West Virginia wind farm through a virtual power purchase agreement (VPPA). This wind farm is expected to generate 125,000MWh of wind energy annually.
Today’s news doubles down on that with the disclosure that Salesforce has signed a second energy agreement, this time with a 24MW new wind farm in Texas which is expected to generate 102,000MWh of electricity annually. When the two wind farms are fully up and running then, Salesforce will be buying 227,000MWh of electricity per annum.
To put this in context, according to its filings with the CDP Salesforce’s total purchase of energy (electricity, fuel, heat, steam, and cooling) in 2015, was just under 152,000MWh. So Salesforce’s energy consumption can grow quite a bit by the time these two wind farms come fully on line in December 2016, and still be well covered by the output of these two wind farms.
If we compare this to a couple of Salesforce’s competitors* –
Microsoft purchases 3,570,438MWh of energy, of which 3,240,620MWh comes from clean energy sources (90.8% clean), and
SAP purchases 918,320MWh of energy , of which 346,885MWh comes from clean energy sources (37.8% clean)
So barring any huge spikes in Salesforce’s energy requirements this year, it looks like they are on track to being the cleanest of the large cloud CRM providers.
In case you are interested in other cloud computing companies purchases of renewable energy, I charted a few of them based on their submissions to the CDP for 2015 – see below
*I tried to find energy and emissions data for Salesforce competitor Workday, but as yet they have not reported their data to the CDP. When they do, I will update this post.
The cost of solar power is falling in direct relation to the amount of solar power modules being produced. With no end in sight to this price reduction, we should soon be in a world where energy is in abundance.
Moore’s Law, the law that says the number of transistors in computers doubles every two years approximately, has an equivalent in solar power called Swanson’s Law. Swanson’s Law states that the price of solar panels tends to drop 20% for every doubling of cumulative shipped volume. This leads to a positive feedback loop of lower prices meaning more solar pv installed, leading to lower prices, and so on. And consequently, as the price of solar panels drops (see top graph), the amount of installed solar globally has increased exponentially (see chart on right).
This law has held true since 1977, and according to the Economist
technological developments that have been proved in the laboratory but have not yet moved into the factory mean Swanson’s law still has many years to run
This positive feedback loop is manifesting itself in China, where in May of this year the National Development and Reform Commission announced that China would target a more than tripling of its installed solar capacity to 70 gigawatts (GW) by 2017.
To put those numbers in perspective, according to the International Energy Agency’s Snapshot of Global PV Markets 2014 report, the total amount of solar PV installed globally reached 177GW at the end of 2014.
And it is not just South East Asia, Brazil and the US this week reached a historic climate agreement that will require.
the US to triple its production of wind and solar power and other renewable energies. Brazil will need to double its production of clean energy. The figures do not include hydro power.
And according to GTM Research, by 2020 Europe will install 42GW and account for 31% of the global solar market.
To fully appreciate the significance of this price, it is necessary to understand that the price of natural gas – which generates 99% of the UAE’s electricity – stands at 9 cents. So, in the United Arab Emirates at least, solar power is currently 65% of the cost of the next cheapest form of electricity production. And its price will continue to decline for the foreseeable future.
So, solar power is cheap (in some cases 65% of the cost of the next nearest competitor), its price is continually dropping, and with 100’s of gigawatts of orders coming into the pipeline, the price reductions may even accelerate.
In this scenario we are headed into a world where solar power rates, for all intents and purposes, approach zero. In that situation, the question becomes, what do we do in a world where energy is in abundance?
Mobile industry consortium GreenTouch today released tools and technologies which, they claim, have the potential to reduce the energy consumption of communication networks by 98%
The world is now awash with mobile phones.
According to Ericsson’s June 2015 mobility report [PDF warning], the total number of mobile subscriptions globally in Q1 2015 was 7.2 billion. By 2020, that number is predicted to increase another 2 billion to 9.2 billion handsets.
Of those 7.2 billion subscriptions, around 40% are associated with smartphones, and this number is increasing daily. In fact, the report predicts that by 2016 the number of smartphone subscriptions will surpass those of basic phones, and smartphone numbers will reach 6.1 billion by 2020.
When you add to that the number of connected devices now on mobile networks (M2M, consumer electronics, laptops/tablets/wearables), we are looking at roughly 25 billion connected devices by 2020.
That’s a lot of data passing being moved around the networks. And, as you would expect that number is increasing at an enormous rate as well. There was a 55% growth in data traffic between Q1 2014 and Q1 2015, and there is expected to be a 10x growth in smartphone traffic between 2014 and 2020.
Fortunately five years ago an industry organisation called GreenTouch was created by Bell Labs and other stakeholders in the space, with the object of reducing mobile networking’s footprint. In fact, the goal of GreenTouch when it was created was to come up with technologies reduce the energy consumption of mobile networks 1,000x by 2015.
Today, June 18th in New York, they are announcing the results of their last five years work, and it is that they have come up with ways for mobile companies to reduce their consumption, not by the 1,000x that they were aiming for, but by 10,000x!
The consortium also announced
research that will enable significant improvements in other areas of communications networks, including core networks and fixed (wired) residential and enterprise networks. With these energy-efficiency improvements, the net energy consumption of communication networks could be reduced by 98%
And today GreenTouch also released two tools for organisations and stakeholders interested in creating more efficient networks, GWATT and Flexible Power Model.
They went on to announce some of the innovations which led to this potential huge reduction in mobile energy consumption:
Beyond Cellular Green Generation (BCG2) — This architecture uses densely deployed small cells with intelligent sleep modes and completely separates the signaling and data functions in a cellular network to dramatically improve energy efficiency over current LTE networks.
Large-Scale Antenna System (LSAS) — This system replaces today’s cellular macro base stations with a large number of physically smaller, low-power and individually-controlled antennas delivering many user-selective data beams intended to maximize the energy efficiency of the system, taking into account the RF transmit power and the power consumption required for internal electronics and signal processing.
Distributed Energy-Efficient Clouds – This architecture introduces a new analytic optimization framework to minimize the power consumption of content distribution networks (the delivery of video, photo, music and other larger files – which constitutes over 90% of the traffic on core networks) resulting in a new architecture of distributed “mini clouds” closer to the end users instead of large data centers.
Green Transmission Technologies (GTT) – This set of technologies focuses on the optimal tradeoff between spectral efficiency and energy efficiency in wireless networks, optimizing different technologies, such as single user and multi-user MIMO, coordinated multi-point transmissions and interference alignment, for energy efficiency.
Cascaded Bit Interleaving Passive Optical Networks (CBI-PON) – This advancement extends the previously announced Bit Interleaving Passive Optical Network (BiPON) technology to a Cascaded Bi-PON architecture that allows any network node in the access, edge and metro networks to efficiently process only the portion of the traffic that is relevant to that node, thereby significantly reducing the total power consumption across the entire network.
Now that these innovations are released, mobile operators hoping to reduce their energy costs will be looking closely to see how they can integrate these new tools/technologies into their network. For many, realistically, the first opportunity to architect them in will be with the rollout of the 5G networks post 2020.
Having met (and exceeded) its five year goal, what’s next for GreenTouch?
I asked this to GreenTouch Chairman Thierry Van Landegem on the phone earlier in the week. He replied that the organisation is now looking to set a new, bold goal. They are looking the energy efficiency of areas such as cloud, network virtualisation, and Internet of Things, and that they will likely announcement their next objective early next year.
We attended this year’s SapphireNow event (SAP’s customer and partner conference) in Orlando and were very impressed with some of the advances SAP and their ecosystem are making in the field of healthcare.
Why is this important?
Healthcare for many decades now has been stagnant when it comes to technological disruption. Go to most hospitals today and you will still see doctors using paper and clipboards for their patient notes. Don’t just take our word for it, in her highly anticipated 2015 Internet Trends report Mary Meeker clearly identified that the impact of the Internet on healthcare is far behind most other sectors.
But this is changing, and changing rapidly. The changes coming to the healthcare sector will be profound, and will happen faster than anyone is prepared for.
And one of the main catalysts of this change has been the collapse in the cost of gene sequencing in the last ten years. See that collapse charted in the graph to the right. And note that the y-axis showing the cost of sequencing is using a logarithmic scale. The costs of sequencing are falling far faster than the price of the processing power required to analyse the genetic data. This means the cost of sequencing is now more influenced by the cost of data analysis, than data collection. This has been a remarkable turn of events, especially given the first human genome was only published fourteen years ago, in 2001.
The advances in the data analytics picking up pace too. In memory databases, such as SAP’s HANA, and cognitive computing using devices like IBM’s Watson, are contributing enormously to this.
To get an idea just how much the analytics is advancing, watch the analysis of data from 100,000 patients by Prof Christof von Kalle, director of Heidelberg’s National Center for Tumor Diseases. Keep in mind that each of the 100,000 patients has 3bn base pairs in their genome, and he’s analysing them in realtime (Prof Von Kalle’s demo starts at 1:00:03 in the video, and lasts a little over 5 minutes).
As he says at the conclusion, two years ago a similar study conducted over several years by teams of scientists was published as a paper in the journal Nature. That’s an incredible rate of change.
IBM are also making huge advances in this field with their cognitive computing engine, Watson. In a recent announcement, IBM detailed how they have teamed up with fourteen North American cancer institutes to analyse the DNA of their patients to gain insights into the cancers involved, and to speed up the era of personalised medicine.
Personalised medicine is where a patient’s DNA is sequenced, as is the DNA of their tumour (in the case of cancer), and an individualised treatment, specific to the genotype of their cancer is designed and applied.
This differs from the precision medicine offerings being offered today by Molecular Health, and discussed by Dr Alexander Picker in the video at the top of this post.
Precision medicine is where existing treatments are analysed to see which is best equipped to tackle a patient’s tumour, given their genotype, and the genotype of their cancer. One thing I learned from talking to Dr Picker at Sapphirenow is that cancers used to be classified by their morphology (lung cancer, liver cancer, skin cancer, etc.) and treated accordingly. Now, cancers are starting to be classified according to their genotype, not their morphology, and tackling cancers this way is a far more effective form of therapy.
Finally, SAP and IBM are far from being alone in this space. Google, Microsoft and Apple are also starting to look seriously at this health.
With all this effort being pored into this personalised medicine, I think it is safe to say Ms. Meeker’s 2016 slide featuring health will look a little different.
Full disclosure – SAP paid my travel and accommodation to attend their Sapphirenow event
Against this background, it is impressive to see to see Equinix, a global provider of carrier-neutral data centers (with a fleet of over 100 data centers) and internet exchanges, announce a 1MW Bloom Energy biogas fuel cell project at its SV5 data center, in Silicon Valley. Biogas is methane gas captured from decomposing organic matter such as that from landfills or animal waste.
Why would Equinix do this?
Well, the first phase of California’s cap and trade program for CO2 emissions commenced in January 2013, and this could, in time lead to increased costs for electricity. Indeed in their 2014 SEC filing [PDF], Equinix note that:
The effect on the price we pay for electricity cannot yet be determined, but the increase could exceed 5% of our costs of electricity at our California locations. In 2015, a second phase of the program will begin, imposing allowance obligations upon suppliers of most forms of fossil fuels, which will increase the costs of our petroleum fuels used for transportation and emergency generators.
We do not anticipate that the climate change-related laws and regulations will force us to modify our operations to limit the emissions of GHG. We could, however, be directly subject to taxes, fees or costs, or could indirectly be required to reimburse electricity providers for such costs representing the GHG attributable to our electricity or fossil fuel consumption. These cost increases could materially increase our costs of operation or limit the availability of electricity or emergency generator fuels.
In light of this, self-generation using fuel cells looks very attractive, both from the point of view of energy cost stability, and reduced exposure to increasing carbon related costs.
On the other hand, according to today’s announcement, Equinix already gets approximately 30% of its electricity from renewable sources, and it plans to increase this to 100% “over time”.
After returning from IBM’s InterConnect conference recently we chided IBM for their aping of Amazon’s radical opaqueness concerning their cloud emissions, and their lack of innovation concerning renewables.
However, some better news emerged in the last few days.
The Whitehouse last week hosted a roundtable of some of the largest Federal suppliers to discuss their GHG reduction targets, or if they didn’t have any, to create and disclose them.
Coming out of that roundtable, IBM announced its committment to procure electricity from renewable sources for 20% of its annual electricity consumption by 2020. To do this, IBM will contract over 800 gigawatt-hours (GWh) per year of renewable electricity.
And IBM further committed to:
Reduce CO2 emissions associated with IBM’s energy consumption 35% by year-end 2020 against base year 2005 adjusted for acquisitions and divestitures.
To put this in context, in the energy conservation section of IBM’s 2013 corporate report, IBM reports that it sourced 17% of its electricity from renewable sources in 2013.
It is now committing to increase that from the 2013 figure of 17% to 20% by 2020. Hmmm.
IBM committed to purchasing 800 GWh’s of renewable electricity per year by 2020. How does that compare to some of its peers?
In 2014, the EPA reported that Intel purchased 3,102 GWh’s, of renewable electricity, and Microsoft purchased 2,488 GWh’s which, in both cases amounted to 100% of their total US electricity use.
In light of this, 800 GWh’s amounting to 20% of total electricity use looks a little under-ambitious.
On the other hand, at least IBM are doing something.
Amazon, as noted earlier, have steadfastly refused to do any reporting of their energy consumption, and their emissions. This may well be, at least in part, because Amazon doesn’t sell enough to the government to appear on the US Federal government’s Greenhouse Gas Management Scorecard for significant suppliers.
What is ResearchKit? Apple’s SVP of Operations, Jeff Williams, described it as a framework for medical researchers to create and deploy mobile apps which collect and share medical data from phone users (with their permission), and share it with the researchers.
Why is this important? Previously it has proven difficult for research organisations to secure volunteers for research studies, and the data collected from such studies is often collected, at best, quarterly.
With this program, Apple hopes to help researchers more easily attract volunteers, and collect their information far more frequently (up to once a second), yielding far richer data.
The platform itself launches next month, but already there are 5 apps available, targeting Parkinson’s, diabetes, heart disease, asthma, and breast cancer. These apps have been developed by medical research organisations, in conjunction with Apple.
The success of this approach can be seen already in this tweet:
After six hours we have 7406 people enrolled in our Parkinson’s study. Largest one ever before was 1700 people. #ResearchKit
I downloaded mPower, the app for Parkinson’s to try it out, but for now, they are only signing up people who are based in the US.
As well as capturing data for the researchers, mPower also presents valuable information to the user, tracking gait and tremor, and seeing if they improve over time, when combined with increased exercise. So the app is a win both for the research organisations, and for the users too.
Apple went to great pains to stress that the user is in complete control over who gets to see the data. And Apple themselves doesn’t ever get to see your data.
This is obviously a direct shot at Google, and its advertising platform’s need to see your data. Expect to hear this mantra repeated more and more by Apple in future launches.
This focus on privacy, along with Apple’s aggressive stance on fixing security holes, and defaulting to encryption on its devices, is becoming a clear differentiator between Apple and Android (and let’s face it, in mobile, this is a two horse race, for now).
Finally, Williams concluded the launch by saying Apple wants ResearchKit on as many devices as possible. Consequently, Apple are going to make ResearchKit open source. It remains to see which open source license they will opt for.
But, open sourcing ResearckKit is a very important step, as it lends transparency to the privacy and security which Apple say is built-in, as well as validating Apple’s claim that they don’t see your data.
And it also opens ResearchKit up to other mobile platforms to use (Android, Windows, Blackberry), vastly increasing the potential pool of participants for medical research.
So I ventured to the conference with high hopes of what I was going to learn there. and for the most part I wasn’t disappointed. IBM had some very interesting announcements, more on which later.
However, there is one area where IBM has dropped the ball badly – their Cloud Services Division, Softlayer.
IBM have traditionally been a model corporate citizen when it comes to reporting and transparency. They publish annual Corporate Responsibility reports with environmental, energy and emissions data going all the way back to 2002.
However, as noted here previously, when it comes to cloud computing, IBM appear to be pursuing the Amazon model of radical opaqueness. They refuse to publish any data about the energy or emissions associated with their cloud computing platform. This is a retrograde step, and one they may come to regret.
This was made more stark for me because while at InterConnect, I read IBM’s latest cloud announcement about their spending $1.2bn to develop 5 new SoftLayer data centres in the last four months. While I was reading that, I saw Apple’s announcement that they were spending €1.7bn to develop two fully renewably powered data centres in Europe, and I realised there was no mention whatsoever of renewables anywhere in the IBM announcement.
Even better than Apple though, are the Icelandic cloud computing company GreenQloud. GreenQloud host most of their infrastructure out of Iceland, (Iceland’s electricity is generated 100% by renewable sources – 70% hydro and 30% geothermal), and the remainder out of the Digital Fortress data center in Seattle, which runs on 95% renewable energy. Better again though, GreenQloud gives each customer a dashboard with the total energy that customer has consumed and the amount of CO2 they have saved.
This is the kind of cloud leadership you expect from a company with a long tradition of openness, and the big data and analytics chops that IBM has. Now this would be A New Way to Think for IBM.
But, it’s not all bad news, as I mentioned at the outset.
As you’d expect, there was a lot of talk at InterConnect about the Internet of Things (IoT). Chris O’Connor, IBM’s general manager of IoT, in IBM’s new IoT division, was keen to emphasise that despite the wild hype surrounding IoT at the moment, there’s a lot of business value to be had there too. There was a lot of talk about IBM’s Predictive Maintenance and Quality solutions, for example, which are a natural outcome of IBM’s IoT initiatives. IBM has been doing IoT for years, it just hasn’t always called it that.
Apple’s recent announcements around renewables and supply chain transparency, put the major cloud providers to shame.
Apple had a couple of interesting announcements last week. The first was that they were investing $848m in a 130MW solar farm being built by First Solar in California. With this investment, Apple enters into a 25 year power purchase agreement with the solar farm, guaranteeing income for the solar farm, and securing Apple’s energy bills for the next 25 years in California. According to First Solar this is the largest agreement in the industry to provide clean energy to a commercial end user, and it will provide enough energy for Apple to fully power its headquarters, operations and retail stores in California, with renewable energy.
For it’s data centers, which hosts Apple’s iCloud, App Store, and iTunes content, Apple uses 100% locally generated, renewable energy. It’s Maiden, North Carolina data centre, for example, uses a combination of biogas fuel cells and two 20‑megawatt solar arrays — the largest privately owned renewable energy installation in the US, according to Apple. And it is now investing another $55 million in a third, 100-acre 17.5MW plant for the facility. You can find details of Apple’s other data centre facilities, and how they are powered by renewables, here.
The second announcement from Apple was the publication of its 2015 Supplier Responsibility Progress Report (highlights here, full PDF here). Apple has been criticised in the past for workers rights violations in its supply chain, so it is good to see Apple taking very real steps, positive, to address this. The amout of detail, the steps taken, and the levels of transparency in the report are impressive.
On underage labour, for instance, Apple’s policy requires that
any supplier found hiring underage workers fund the worker’s safe return home. Suppliers also have to fully finance the worker’s education at a school chosen by the worker and his or her family, continue to pay the worker’s wages, and offer the worker a job when he or she reaches the legal age. Of more than 1.6 million workers covered in 633 audits in 2014, 16 cases of underage labor were discovered at six facilities — and all were successfully remediated.
Comparing Apple’s cloud offerings to actual enterprise cloud players (or any cloud players, for that matter), you see there’s a yawning chasm in terms of transparency, reporting, and commitment to renewables.
Of the main enterprise cloud players:
Microsoft publish their Citizenship Report here [PDF]. And while it is a decent enough report, it doesn’t go into anything like the level of detail that Apple does. On page 53 of this report Microsoft mention that 47% of the energy it purchases is renewable. It does purchase renewable energy certificates for the other 53% so it can report that it is carbon neutral.
Google doesn’t produce a corporate sustainability report. Instead it has this page which outlines some of the work it does in the community. Information on Google’s energy breakdown is sparse. What is published is found on the Google Green site, where we find that although Google has many investments in renewable energy, and Google has been carbon neutral since 2007, Google’s actual percentage of renewables is only 35%.
IBM has a good history of producing corporate reports (though it still hasn’t published its report for 2014). However on the energy conservation section of IBM’s corporate report, IBM reports that sources 17% of its electricity came from renewable sources in 2013. However, they go on to note that this does not include the energy data of Softlayer – IBM’s cloud platform.
And finally, Amazon, who have arguably the largest cloud computing footprint of any of the providers, is the worst performer in terms of reporting, and likely in terms of emissions. The only page where Amazon mentions emissions, claims that it has three carbon neutral regions, but fails to say how they have achieved this status (or whether they are third party audited as such). The same page also claims that “AWS has a long-term commitment to achieve 100% renewable energy usage for our global infrastructure footprint” but it fails to give any time frame for this commitment, or any other details on how it plans to get there.