Tim Kham from ANNA Systems, LLC: “Russian IoT market is more likely to be developing than innovative”
Key problems of the Russian IoT market are common usage of overseas sensors and technologies as well as poor B2C. That is an opinion of Tim Kham, a participant of the discussion dedicated to the evaluation of consumer and corporate IoT solutions, Founder at ANNA Systems, LLC that developed an HPC service called A-Platform. In his interview for our event, Tim told about the state of things on the Russian IoT market and in the sphere of high-performance computing as well as shared his opinion on home IoT-powered products.
Interviewer: VI international Internet of Things forum (IOT).
Respondent: Tim Kham (Т. K.).
IOT: Hi, Tim. How well is the Russian IoT market developed?
Т. K.: Hi. A significant increase in the interest in IoT can be observed in Russia. However, it should be noted that apart from prospects this industry opens, there is a range of challenges to focus attention on:
- It is business owners who are more interested in Russian IoT technologies than individuals. I guess the point is the economic situation. To exemplify, even a small Russian B2C market is rife with overseas companies. Their products are verified, better advertised, and sold at multiple stores. I believe that the current market situation demonstrates that B2B solutions are the best way of development in the IoT sphere.
- The Russian IoT market is more likely to be developing than innovative. Sensors are installed at plants, built in electricity meters, vehicles, and refrigerator trucks. Moreover, smart databases and data centers are created, smart grids are modernized to fully comply with the standards for modern electrical grids. In addition, surveillance cameras are installed, even elevators in multiple blocks collect statistical data and can be managed remotely. However, you can’t fail to notice that the majority of domestic developers either produces a carbon copy of the Western technology stack or buys ready-made stack elements. For instance, famous system STRIZH leverages LoRAWAN (including transceiving chips on both sides). NB-IoT provided by mobile operators is entirely based on foreign software and hardware. The manufacturing process is mainly outsourced to China, chips (microcontrollers, transceivers, sensors) and relevant SDKs usually come from overseas. Even the development of systems for machine learning and data collection either open-source software or commercial off-the-shelf stack elements used that are simply interconnected within a source code that mostly defines the system interface.
- Home software stacks and hardware usually do not meet the necessary (and even provided by the manufacturer) operating, stability, and safety requirements.
- Another problem of commercial IoT distribution is executive staff at enterprises. Top managers are not interested in any sort of innovations unless they generate immediate payoffs.
I would recommend developers of IoT-based solutions thinking about the possible product expansion to foreign markets right away. It would allow to diversify the target market as well as provide business with new possibilities and advantages.
IOT: As an experienced developer, could you compare domestic IoT solutions with European and American ones?
Т. K.: We often lack quality. Speaking about consumer IoT solutions, their quality leaves much to be desired. In terms of industrial IoT, the conditions are a lot better: there are industrial standards, competition as well as working infrastructure.
The reason is that commercial IoT in fact emerged at the end of the last century in the form of expensive industrial networks at factories, inside of tankers, and other complex infrastructure objects. What is more, GPS and GPRS led to the appearance of trackers (in many countries, the insurance cost depends on the presence of a tracker with sensors in a vehicle, that is why every car has a tracker) and control systems (aimed at the abovementioned refrigerator trucks and elevators in residential areas). There is no shortage of examples. It’s hard to compare the Russian consumer IoT segment with the Western one for the sharp difference in development.
IOT: In your point of view, what spheres does IoT penetrate most of all and why?
Т. K.: Speaking about Russia, these are no doubt industry, nuclear power, and logistics. Everything depends on money and staff who have to understand how to use cutting-edge technologies to reduce expenses.
In Europe and America, IoT has already become part of the abovementioned spheres, now it’s time for the consumer segment. However, there also arise problems: in fact, fitness trackers and cheap surveillance cameras are the only to have gained popularity. Voice assistants in the form factor of speakers such as Alexa and Google Home as well as Apple speakers also grew popular. The majority of people use them to listen to music with voice-activated control or aimlessly chat with a bot in one’s free time.
It is undoubtedly that IoT locks, IoT sensors for apartments and houses, IoT-powered fridges, smart air conditioners, and other stuff from the universe of consumer electronics also found their target market. However, they are mostly used by geeks and representatives of the developing company.
For instance, companies buy smart fridges for their kitchens just to maintain a façade of success and innovation hub (they usually are a perfect match for the general hi-tech design of the office). Nonetheless, smart functions are used once or twice just for fun.
Lack of standardization and compatibility deters IoT from the real promotion in the consumer segment: all manufacturers try to create a rigid connection between users and services thus maximizing the current profit.
Let’s crack on smart houses: newly-bought devices are rather incompatible with the existing system. The absence of an electrical grid in necessary places, no integration with a single controller, apartment renovation is needed to install smart devices. Creators of such ready-to-operate homes deal with the challenges in a corporate way: all systems are installed, wiring is conveyed, and then it’s time for renovation. The implementation of such solutions is expensive, which prevents their mass distribution.
Taking into account the current development of wireless technologies and batteries as well as societal digitalization on the whole, I think that the IoT market will demonstrate big changes at all levels within the nearest ten years.
IOT: Could you describe the current situation of the market of high-performance computing?
Т. K.: We can currently observe the following state of things on the market of HPC and HPC Cloud:
- There is a high price barrier for entering the market on conditions stipulated by traditional players (equipment and software vendors). For instance, the price of hardware needed to assemble a small computer cluster (200-300 kernels) ranges from $0.5 to 2 m. One use license for application software (for instance, ANSYS Mechanical Enterprise) with a special HPC app for code parallelization into 16 kernels would cost about $200 000. Traditionally, to solve a task in the best way and process received data, you will need additional software suites for the price exceeding an entrance barrier. Cluster maintenance also requires expenses. You cannot just install such a cluster in a cabinet or in the kitchen. Mere power consumption of a cluster with 200 kernels would averagely cost about $10 000 per month.
- Infrastructure of HCP cloud services is underdeveloped because of the lead maintained by the conventional business model. Equipment and software vendors against large corporations and state institutes.
- Cloud services are not user-friendly. They require much reconciliation with managers and manual work for data loading. That can be explained by the presence of heavyweights from the traditional segment of cloud services (Google, Microsoft, Amazon) that ignore the peculiarities of high-performance computing services.
- Currently, users see the majority of high-performance computing services as a non-user-friendly environment requiring profound knowledge in specific areas such as system administration and programming.
IOT: How did the idea of ANNA Systems, LLC foundation spring to your mind? How did you analyze the market before starting a project?
Т. K.: We teamed up in 2012. Initially, the company was focused on neural network algorithms aimed at robotic machine vision. They require ongoing training in order to boost adaptability to ever-changing environmental conditions.
Ongoing systematic training of such algorithms requires high capacity. At that moment, we couldn’t afford a computing cluster, install special commercial software, and spend some money on its maintenance monthly.
There appeared the need for services for remote high-performance computing with access through a usual PC. It appeared that services working on 24/7 basis can hardly ever be found.
From a commercial point of view, capacities held by large state organizations and corporations were not suitable in terms of organization and bureaucracy as well as client-centeredness.
At some point, we turned out to be not the only one to suffer from it. By 2014, a potential segment in the form of small and medium business has already been formed. It was presented by small teams of developers, engineers, builders, designers, startupers with the need for computing capacities and well-trained staff for processing data of various kinds: from rendering and CAD reports to machine learning and Big Data.
On the other hand, heavyweights and state institutes frequently have clusters and supercomputers loaded ~70% at best. Even database equipment is overloaded (as of 2017) by ~70% in Russia and ~95% in the USA (save for cloud services of IT giants, this figure will slash to 70%).
We think that this leads to a sort of digital divide: many companies are in fact cut off new technologies while others have unprofitable capital downtime. That is why the mission of our company is the creation of the infrastructure of cheap access to technologies (SaaS) and calculations (IaaS) for all the interested Web users across the globe.
Since 2015, our team has been engaged in learning HPC technologies (High-Performance Computing) and their adaptation to Internet users (Cloud HPC).
IOT: Please, provide more details on your product called A-Platform. What tasks does it solve and who is its target audience?
Т. K.: A-Platform intends to build the global infrastructure allowing every Internet user to apply high-performance computing.
Key points of our plan:
- considerable reduction of these services’ cost;
- a user-friendly online platform with a convenient interface that adapts to search queries;
- sharp drop in an input threshold for use of high-performance computing services;
- self-developing ecosystem of HPC services for all market segments, including the technologies and tools for the development of adjacent markets: IoT, Big Data, Virtual Reality, Artificial Intelligence.
Specific steps we intend to take in order to fulfill our plans:
- establishment of our own high-performance computing infrastructure;
- development of software allowing any individual to obtain access to high-performance computing services online through a web interface, APIs, or mobile apps;
- development of the means of effective connection of powerful high-performance computing capacities to global computing infrastructure A-Platform;
- development of APIs allowing to use HPC for the advance and mass inclusion of IoT, including micropayments;
- development of additional components of the HPC ecosystem, including the marketplace of algorithms, models, and apps.
We utilize A-Platform to consolidate distributed and disseminated HPC resources of partners, allow to commercialize clients’ computing services, and provide access to the ecosystem and a knowledge base to resolve typical (CAD, CAE, CAM, Rendering, 3D modeling, etc.), highly specialized (CFD, Deep Learning, Machine Learning, Big Data, IoT, etc.), and specific tasks (Data Science, Artificial Neural Network, etc.)
IOT: Being Founder at A-Platform, tell us please how it correlates with the Internet of Things technology?
Т. K.: A-Platform provides two essential things: computing resources and machine learning algorithms. The IoT market cannot function without these aspects.
IOT: You are a participant of the discussion dedicated to theevaluation of consumer and corporate IoT solutions from the investors’ point of view. What issues would you like to feature?
Т. K.: Personally speaking, processing of data collected by IoT-powered sensors would be of the biggest interest. Machine learning such as SaaS as well as provision of computing capacities such as IaaS.