Matrix AI — 2019 Year in Review – by Steadydee

1 ofD3bF71PiqdGODiNjyQiQ

This is a copy of a great article published here:


Matrix AI is an AI-based blockchain headquartered in Beijing, China that was launched in late 2017 on the tail end of the infamous cryptocurrency bull run.

Members of the Matrix team including Tim Shi, Bill Li, Steve Deng, and Owen Tao (left to right)

Fast forward to January 2020 and and two years have passed since Matrix AI began their quest to create a next generation AI-infused blockchain. In this article, I take a look at their progress toward this goal and specifically, what they have accomplished over the past year.


In reviewing the Matrix AI team’s progress for 2019, I cover the following areas:

  • Preface
  • Background
  • Infrastructure
  • Projects
  • DApps
  • Partnerships
  • Events
  • Community Activities
  • AMAs
  • Matrix-Produced Videos
  • Media
  • Status for 2019
  • Conclusion


In general, a lot of has happened over the last two years in the crypto space since Matrix AI first launched, most conspicuously a bear market that has seen prices fall precipitously, up to 99% for many crypto projects. One could argue that this correction was more than necessary after rabid speculation that led to a parabolic increase in prices for a period during the 2017 bull run.

A bear market for cyptocurrency prices

As a result of the bear market and dramatic fall in market cap for altcoins, many people have written off crypto projects as “shitcoins” that were really just a grab for money. Certainly there have been plenty of scams in the space and the SEC in the United States has pursued court action against a number of companies.

However, even though some unscrupulous companies took advantage of frothy investors and a lack of regulation, a closer look at the market reveals that there are still many worthy and exciting projects that are advancing blockchain technology and offering real value-add use cases for the economy.

This is where a fundamental review of this sort often comes in handy, identifying promising crypto companies based on the progress they have made toward stated goals. This is especially important when trying to review a project still under development because in most cases they can’t be evaluated like a traditional company (no earnings), on price (too much speculation) or based on measures of network activity (no transactions).

While investing in what is still a relatively nascent industry is very high risk and speculative, projects that have been able survive and even thrive in the 2018–2019 crypto bear market are worth looking at for the potential impact they could have in what is one of the most promising advances in technology that we have seen in a generation.

With that said, and without further ado, let’s take a look at one such project, Matrix AI, and see what they accomplished in 2019.


For Matrix AI, 2019 was the year that saw the company’s 2018 investment in development pay off in a big way with the launch of their mainnet. For a blockchain, the mainnet is the foundation upon which everything else is built and it has been no different for Matrix AI.

In the case of Matrix AI, building their mainnet was no small task. Instead of marginally improving an existing blockchain as many crypto projects have done, they decided to start fresh by completing reinventing Proof-of-Work (PoW) consensus. Instead of tying up the entire network’s computing resources in a redundant hashing exercise, they envisioned a blockchain where only a handful of randomly selected computers would perform mining activities in order to secure the network.

In redesigning traditional blockchain consensus, they were able to free up the majority of the network computing power to run AI algorithms, basically allowing the network to act as a worldwide, distributed AI cluster.

Furthermore, by combining the approach of randomly selected miners with a layer of Proof-of-Stake (PoS) validators, they addressed the 51% attack associated with PoW blockchains (the reason BTC uses so much energy) and the Nothing At Stake attack typically associated with PoS blockchains.

Matrix AI has designed a blockchain that can do everything that Ethereum can do (e.g., smart contracts, issue tokens, DeFi), while being arguably more secure, and at the same time reallocating most of the processing power of the network to run AI algorithms. One could argue that they have a more complete vision of blockchain as a “world computer” than Ethereum.

Envisioning a new blockchain model and actually building it are two different things. Delivering their mainnet in a little over a year was quite an accomplishment considering that a similar effort to redesign Ethereum may take years to finish according to some projections.

By all measures the mainnet has been a success with over 1200 mining nodes and 65 validator nodes running globally, blocktimes consistently under 12 seconds, and mining, pooling and staking working without a hitch.

Mining and Validator Nodes

This has set the stage for the launch of AI processing on the mainnet. A custom-produced AI mining machine called “Apocalypse” and GPUs are set to be added to the network in February 2020 with customers expected to be onboarded shortly thereafter.

Integrating AI into the core of the blockchain enables a whole host of advantages over other blockchains such as improving performance, security, and privacy, but at its core, the ability to offer cheap computing power to run AI algorithms is the unique value proposition of Matrix AI.

This is especially relevant at a time when the amount of computing power required for running AI models is increasing at an exponential rate, even faster than Moore’s Law, threatening to put the cost of AI out of reach for many companies. This recent tweet from an AI researcher demonstrates the problem:

The cost of AI computing is likely to increase if current projections are any indication. A PwC study found that AI’s contribution to the global economy by 2030 could be a whopping $15.7 trillion. To put that in perspective, that total would exceed the current combined economic output of China and India.

Estimates such as these are not out of line. AI is likely to be the most transformative technological advance of our lifetime. AI has the potential to disrupt every sector of the economy and companies that fail to adapt and utilize AI will likely be put at a disadvantage.

This raises the question of whether AI will be democratized or if only the largest corporations will be able to afford sufficient access to AI.

Projects like Matrix AI have the potential to help democratize access to AI by addressing the following issues:

  • The high cost of computing which limits access to AI.
  • The centralization of data, today controlled by the Googles of the world.
  • A lack of the talent needed to develop AI algorithms.

By providing low cost computing, open and secure access to data, and a marketplace of algorithms, one could say they will help make AI more accessible, something worth noting when a handful of the world’s most powerful companies are threatening to monopolize this game-changing technology.


With a major focus on development and building out the mainnet, it’s not surprising to see that Matrix AI expended a significant amount of work on infrastructure. At the center of this effort was a release of the testnet, followed by the mainnet, and afterward by a number of important updates to the network and GMAN client, along with provisions for staking and pooling.

Other areas addressed this year included the release of the Matrix 2.0 Green Paper, and infrastructure outreach including a developer portal and an ecosystem platform. The stage has now been set for the introduction of GPU and the running of AI algorithms directly on the network.

Below is a full list of infrastructure accomplishments and their source:

  • Testnet Launch — A testnet allowing the community to test the network using test tokens.
  • GPU testing — The testnet was used to test GPU and an AI algorithm.
  • Mainnet Launch — The most important event of the year as Matrix AI put the foundation in place for their AI ecosystem. The Hybrid PoW/PoS consensus model, dubbed HPoW frees up the majority of the network’s computers to process AI algorithms.
  • GMAN Update — Fixed and flexible staking introduced.
  • GMAN Patch — Designed to increase stability of the network.
  • Participation Incentives Update — Incentives and elections tweaked on the network.
  • EasyGMAN with snapshot files — Snapshots save significant syncing time.
  • EasyGMAN —A new tool that enables the launching of a Masternode with one click.
  • Blockchain Data Syncing Update — Snapshots added to the update.
  • Major Network Update — Mainnet AI mining and a new block creation method, alongside changes to the mining process, the election algorithm, the reward distribution method and a penalty policy for inactivating miners.
  • Blockchain Syncing Update — Reduces the size of the overall chain and syncing time.
  • Joint Mining Introduced — Pools for verification nodes.
  • Apocalypse AI Mining Machine Introduced — A custom machine that supports data mining, AI training, 3D rendering, meteorologic analysis, biological computing and clinical diagnosis.
  • The Matrix Ecosystem Platform Launched — An all-in-one platform for the community to access Matrix DApps, Matrix Wallet, AI-powered tools and many other functions.
  • Matrix Developer Portal Launched — a platform providing developers with the tools and data needed to build apps
  • Matrix 2.0 Green Paper — Describes the Matrix 2.0 architecture and ecosystem and how its approach to data, computing power, and AI models is aligned with the “Fourth Industrial Revolution.”
  • Matrix Wallet App for Android
  • Stress Test — In a public stress test, the Matrix AI network reached a top speed of 12,499 TPS under real world conditions.
  • Token Swap — No deadline was given and Matrix continues to provide support to users who want to swap ERC-20 tokens.
  • Silicon Valley Office — Matrix successfully registered for a business license in Silicon Valley.


In parallel with infrastructure development and led by Chief AI Scientist Steve Deng’s AI group, a significant amount of effort has been devoted toward designing AI-based solutions for industrial and medical sectors within China. As a part of this effort, the Yangdong Artificial Intelligence Research Institute was created to be a focal point for the team for AI research and development.

One area of focus is “predictive maintenance” which uses AI to anticipate maintenance and safety issues and improve downtime with large industrial machines and high-speed rail, the latter being an area where Deng has significant corporate and academic research experience and national research funding.

A second major area of focus is using AI to diagnose illnesses such as cancer and fractures. Research studies have shown AI systems often have higher levels of accuracy in diagnosis when it comes to reviewing medical scans. Deng has also received national funding for this research and has established relationships with major hospitals in China.

Predictive maintenance and processing of patient data is in the top 10 AI use cases

Another observation from having reviewed a number of Matrix AI articles is that the team is focused on important issues of technology integration. They are developing solutions for integrating the Matrix AI blockchain into existing industry infrastructure such as edge devices (embedded devices, IoT sensors, GPS, RFID, etc), corporate enterprise systems (ERP), cloud infrastructure, and communication systems (5G). This research should expedite the process of onboarding and enabling industry to fully utilize the promise of AI on blockchain.

To learn more about these projects, see the Matrix-published articles listed below:

  • Yangdong Artificial Intelligence Research Institute — This institute named for Matrix’s Chief AI Scientist was created to combine blockchain, AI, IoT, and big data in order to provide next generation solutions to industry.
  • Matrix Enterprise Cloudchain — Aims to support the operations and maintenance of several industrial production applications by providing distributed AI computing resources and models
  • IoT Cloud and Intelligent Warehouse Solution — An IoT cloud platform that aims to further the development of “Intelligent Logistics” and “Warehouse Management” including power monitoring, dangerous object detection, fire detection, and security monitoring.
  • Predictive Maintenance — Predict faults and maintenance issues for industry equipment in areas such as high speed rails. One such platform is being tested with 48 locomotives in Kenya and 201 locomotives at the Fengtai Rolling Stock Maintenance Bay in ChinaProfessor Deng has worked for a number of years with with railway giant, China Railway Rollingstock Corporation, and has procured national research grants in the area.
  • Intelligent Healthcare — An accessible intelligent platform aiming to provide direct access to medical information and more accurate treatment of patients.
  • Intelligent Cloud Monitoring Service — Being designed to measure wind power and detect its impact on mechanical equipment and can monitor the health of large-scale structures such as bridges and dams.
  • AI-powered Liveness Detection System — Matrix AI is developing an AI-powered liveness detection system which, with the help of 3D cameras, can accurately and consistently recognize a still human face.
  • Smart platform being developed as part of a Smart City Firefighting and fire-prevention project.
  • Cancer diagnosis and rib fracture projects with Shanghai Pulmonary Hospital (affiliated to Tongji University) and the Huadong Hospital (affiliated to Fudan University).
  • EpiCognition — A service offering to connect traditional industry with AI solutions powered by small data.


Partnerships in the crypto space have received a bad rap over time, mostly because of projects artificially generating partnership announcements in order to boost price, even in cases when the likelihood of actually working together may be very low.

Partnerships or “strategic agreements” to explore ways of working together are generally very tenuous things when the product (blockchain) is still under development. Circumstances change over time and by the time the product is ready, companies may have headed in a completely different direction.

In reviewing partnership announcements by Matrix AI in 2019, it appears that some may work to significantly benefit both parties (Theme Tech and Bitgrit) while others, even if formed originally with the best intentions, appear not to have worked out based on currently available information (MBC Project).

Theme Tech — A Beijing-based company and the official service provider of China Air Traffic Management Bureau that will utilize the Matrix AI network for AI-based wind and weather forecasting at airports.

AI-powered commercial weather data platform

Bitgrit — Bitgrit is an established community of data scientists and marketplace of AI algorithms which may be a good fit for the low-cost AI computing resources of the Matrix AI network.

DAIA — DAIA is an alliance of diverse organizations working on decentralized AI and founded by Singularity NET’s Ben Goertzel and AI Decentralized’s Toufi Saliba.

MBC Project — An intelligent shipping project proposing to use AI and blockchain to create greater efficiencies in the shipping industry. According to announcements in the MBC English Telegram group, MBC and Matrix AI have gone their separate ways although it was announced that Deng is currently providing AI consulting for MBC.

Intelligent Electrical Charging Chain (IECC) — The IECC is developing a blockchain electric car charging ecosystem that would reward users of the system with tokens and while performing Matrix AI Network mining and deep-learning tasks while idle.


Once the infrastructure is built, industry solutions developed, and partners are in place, AI projects will likely take the form of Decentralized Apps (DApps) available through the Matrix AI Ecosystem.

Unlike some blockchains that have rapidly pushed low-value DApps such as gambling and gaming in order to drive up the number of onchain transactions, Matrix has indicated that they want to develop a handful of high-value projects and DApps for clients to set the standard going forward in terms of AI applications.

Even though the ecosystem is not fully ready for DApps, Matrix AI has been actively developing and releasing a number of fully working DApps to the public. Although it is likely that at some point these DApps will be integrated into industrial and medical settings, allowing these to be consumer facing initially provides a good overview as well as public testing of what Matrix AI is working on.

The following is a list of launched DApps:

DApps to be launched shortly:


Representatives of the Matrix AI team were busy attending a variety of events over the past year in a number of countries in the region including China, Japan, Turkey, South Korea, and Vietnam as well as supporting meetups held by region community groups.

Bill Li, Matrix CTO attended a China National Blockchain Standards meeting Shenzhen, China, November 27.

Community Activities

Matrix AI maintains an active outreach to Telegram communities around the world including in the U.S., Korea, Japan, Vietnam, India, Russia, Turkey, Brazil, Italy and Spain. The following is a list of some of the main activities that occurred in the English Telegram group including some major contributions from community members.

Matrix Masternode Docker Setup — An amazing design for implementing multiple nodes on a server using Docker Containers, now the defacto setup for mining nodes on the Matrix network, created by community member Mike Pence @Pencekey.

The Docker Container System for Nodes


Throughout the year, the Matrix AI management team made themselves available to answer questions from the community in the Ask Me Anything (AMA) format.

Matrix-Produced Videos

Matrix AI produced 17 videos over the course of the year, mainly addressing technical questions and how-to’s for users in the community.


Many blockchain projects in the space have dedicated a significant amount of resources to marketing and promotion. Perhaps because the Matrix AI team is mainly made up of researchers and scientists, or because their strategy emphasizes building a product first before promoting it, marketing has been less of a priority. Nevertheless a surprising amount of media activity involving Matrix AI was recorded in 2019.

For example, there were a good number of references to Matrix AI in the media (16 articles) and there were a number of direct interviews with Deng (9). Even though interest in crypto may be at a low point due to the bear market, Matrix AI was still able to attract interest in both the English and Chinese speaking press during the year.

In addition, Matrix AI was incredibly active writing articles on Medium, having published 78 articles over the course of the year. They also maintained a strong presence on Twitter with nearly 400 tweets or retweets.

Interviews with Steve Deng

Articles Referencing Matrix

Stats for 2019

Number of Tweets or Retweets by Matrix — 396

Number of Medium Articles — 78

Number of Matrix-Produced Videos — 17

Media Interviews with Steve Deng — 9

Media Articles Citing Matrix AI — 16

Number of Events — 13

Number of AMAs — 5


When Matrix AI announced their ICO in 2017, they were an unknown entity out of China with a grandiose vision of how AI could be used to reinvent blockchain. Even though Matrix AI put together a team of accomplished researchers, scientists and entreprenuers, investors in the ICO were taking a significant risk on whether these ideas could be turned into a real, working product that would satisfy a demand in the marketplace.

Although the question of whether Matrix AI will realize their goal of providing cheaper and decentralized AI resources to the marketplace is still unanswered, much of the initial risk surrounding the project has been mitigated. After an extremely productive 2019, Matrix AI is on the verge of offering AI services to companies and governments, possibly in the next few months. In addition, the stage is set for their planned AI marketplace with incentives for sharing data, AI algorithms, and computing power in one integrated ecosystem.

The timing for Matrix AI could not be better. China has made a commitment at the highest levels of government to lead the world in blockchain and AI. The country is actively promoting blockchain projects at all levels of society. Combined with the accelerating global demand for AI services and the rising costs associated with AI processing, Matrix AI is facing a perfect storm of opportunity in the market.

If you liked the story, please share and give it some claps!

For more AI and crypto related tweets, catch me on Twitter.

1 Comment on "Matrix AI — 2019 Year in Review – by Steadydee"

  1. This is so great article! Great job SteadyDee!

Leave a comment

Your email address will not be published.


Scroll Up