Primary clients for this platform is AI system developers. Real market size for this platform is not as huge. Researchers are already involved in another AI ecosystem.
No information about current stage of platform development. Stages of platform development is connected with attracted money. All working examples was made in Russia, despite of the fact, that nominally company is based in Tallinn.
No first revenue from clients. The miner will receive reward in NeuroTokens, but this function is not developed yet. Neuromation Lab is crucially depends on the quality of the models and industrial solution, however, nowadays we could not predict that.
Projects goal is not matched with the financial projections of the project. 55% of projected revenue based on unconfirmed assumptions.
Team works remotely Almost all team members involved in project part-time
Platform have to generate a huge amount of labeled data and as a result, it provides an opportunity to train ML models and creates industrial solutions. Market of industrial AI solution is already huge and has double-digit growth rates yearly. The potential demand for image recognition alone in the retail industry is enormous. It amounts to more than 40 billion images per year, according to ECR research of 72 of the biggest retailers and suppliers. Also, other implications of AI models will completely transform different industries. According to Gartner, by 2020 85% of customer interactions in retail will be managed by artificial intelligence. As a result, the leading platform for decentralized computation has great perspectives. Also, a quantity of cryptocurrency miners is growing, their profits are declining. As a result, it would be easy for the platform to attract miners and their computational power. The trickiest part is attracting consumers and researchers on this platform. Researchers are already involved in AI ecosystem, where they can share and compare their results. Also, corporations usually have special plans for researchers and provide them computational power with the substantial discount. At the same time, scientists do not ready to pay a fortune for unique labeled data sets. Final consumers of AI industrial solution usually want to get an entire solution, rather than labeled datasets of images. As a result, primary clients for this platform is AI system developers. This is completely another size of the market.
Neuromation main product is a platform on Ethereum blockchain for decentralized building a library of synthetic data using the computing power of private and commercial mining service providers. Using the sets of these data, Neuromation can effectively train models of neural networks. At that moment, there is no information about current stage of platform development. There is few video examples, however, this videos provides only general principals of platform functions, there is no actual examples of working prototypes. Team hopes, that it can develop system, where current cryptocurrency miners could use their decentralized computation power to solve real world problems and generate synthetic data for ML models trainings. For some reasons, stages of platform development is connected with attracted money. Moreover, team has ambitious goal to attract more than 100 000 GPU to their platform in Q3 2018, however, they do not have any strategy to attract enough consumers to fulfill such computation power. Also, it means, that computation power will grow up 100 times within only 3 quarters, but at the same time, projected cash flows is not grow that much fast. Key feature of the project is Neurotokens (NTK), it is the primary exchange mechanism enabling synthetic data generation, distributed model training, data labeling and other AI services. It is also assumed, that it would be introduced in Q3, however, now we could not see even prototype or code of the platform. Second product is Neuromation Labs. It is AI industrial solution developer with SaaS model. It is assumed, that large sets of synthetic data would provide an opportunity to create leading AI solution in different industries. Currently, company has completed project in retail industry. However, all working examples was made in Russia, despite of the fact, that nominally company is based in Tallinn, chairman and investor is based in USA, founder and CTO is in Kiev, CIO is in Tel Aviv. Also, code is not available, only few pictures on site shows the whole results of this project.
The project has 2 main parts. First is the platform for generating and exchange synthetic data sets, developing and training AI models within members of the ecosystem. The second part is Neuromation Lab, they assume, that it would earn money by selling an API (SaaS) or data to the industry partner. In fact, first part works just as a marketplace. The biggest problem for such business model is attracting both sides of its clients. Usually, one side has excessive capacity or supply and want to find new ways of generating revenue. The same thing happened in that case. Team states, that they have the great proposition for the crypto-currency miners that could attract their computing resources and help resolve the client's problem. However, it is not supported by any facts. There is no first revenue from clients, it is not proved, that clients are ready to pay for such distributed miners substantially more than they currently earn on their prime business. The whole process is designed as an auction. When a Neuromation task is available, any eligible node can bid to participate. If a node wins the bid, it will switch computing power from mining Ether or other coins to a task on the Neuromation Platform. The node will generate synthetic data or train a deep learning model. As its reward, the miner will receive NeuroTokens (NTK). Once the task is complete, the Neuromation Node will exit and the miner will proceed to mine cryptocurrency. Initially our NeuroTokens (NTK) will be extending Ether, but later on, we plan to migrate to our own blockchain, once the Neuromation Platform economy matures. The second part of their business crucially depends on the quality of the models and industrial solution, however, nowadays we could not predict that. Prototypes are not available for potential tokens buyers.
It is strange, that project does not have minimum requirements of attracted money. It seems, that project is close to seed stage, as a result, hard cap 30 mln USD (60 000 000 NTK = 60 000 ETH = 38 520 000 USD) seems extremely high. The project has 2 main revenue streams. First is a commission from transactions on the platform. The commission is from 5% to 15%, it depends on the type of service received through the Neuromation Platform. However, the team does not disclose exact commission from each type of service. Team project around $71 million USD in gross transaction volume on the Neuromation Platform for 2018. Also, they expect, that platform has to grow from 3x to 5x by 2022. That is mean, that they expect, that around 200-350 mln USD would be the gross transaction at the platform. At the same time, they give the forecast of the market growth from 70% to 150% yearly up to 2025. That is mean, that even they do not believe, that company is going to became a leader at that market. Also, the estimated price for creating data generator and model sale states at 20 000 USD, which seems extremely high. This price is not proved by any market analysis or comparison. Also, they expect to sell 1000 data generators and models in 2018, however, at the same time, their adviser told in the interview, that models do not cost a fortune, labeled data is the most important resource. All in all, revenue from model sale and creation of data generator is more than 55% of their projected revenue. The second revenue stream is Neuromation Labs, they expect to partner with industry leaders to spearhead AI solutions with them. Neuromation makes money by selling an API (SaaS) or data to the industry partner. There are no exact projections of revenue from this stream. All in all, stated goal “One of the main goals of the Neuromation platform is to become a center for artificial intelligence services for international business” is not matched with the financial projections of the project.
At the first glimpse, company has a strong team. However, in fact, it seems, that all members of team work remotely. And also almost top managers are involved in other projects, which mean, that none of them is not fully dived into Neuromation. Further, is exact information about key team members. First of all, Fedor Savchenko - CTO is also involved as Lead Designer at Biblio-Globus, Moscow based company and as art director and partner of Thrive.io. That is mean, that he is not fully involved in the project and work remotely. Secondly, Sergey Nikolenko, CRO, is also an assistant professor at NRU - Higher School of Economics in St. Petersburg, an adjunct professor at St. Petersburg Academic University of the Russian Academy of Sciences and researcher at Steklov Mathematical Institute. He also could work only remotely and it seems, that Neuromation is not his first priority. Constantine Goltsev is chairman and investor at Neuromation. He has a great experience, however, also involved in another project. He is also CEO as SolidOpinoin.com, which is based in USA. Next is Denis Popov, CIO of the project. He is also CTO and Founder of Coding-Rocks, Kyiv based company. Esther Katz, VP communication, does not even state, that she is involved in this project on LinkedIn, but at the same time, she states other projects and self-employment. Evan Katz, Chief Revenue Officer is located in Tel Aviv and states, that he has started his work in Neuromation only in December 2017. The same situation with advisors. None of them states, that they are involved in a project and they also spread over all world. In a nutshell, I completely cannot imagine, how such team could create a market leading company.