Author | Song Jiating Editor | Luo Lijuan
To this day, the industry still has different opinions on whether NVIDIA (Chinese name "NVIDIA") is a chip company or an artificial intelligence company.
But NVIDIA founder and CEO Jensen Huang has said on multiple occasions: "NVIDIA is an artificial intelligence company."
The change began a few years ago, and artificial intelligence has become a global Rising, the company seized the opportunity. Relying on its technical accumulation in graphics processing units (GPUs), NVIDIA quickly transformed from a graphics chip company to an AI platform builder and achieved great success.
In the past few years, Nvidia’s stock price has more than doubled, and its market value once exceeded 100 billion U.S. dollars, making it a hot artificial intelligence company in the world.
From games, autonomous driving to robotics and other popular areas of AI, NVIDIA is everywhere.
In the medical industry, some institutions predict that the valuation of AI medical care will reach US$6.6 billion by 2021. Although the technical threshold in this field is high and difficult to implement, any player claiming to be an artificial intelligence company is not willing to miss this big piece of cake.
Having conquered the hard core of technology, NVIDIA has finally ushered in the product launch period. At the 2019 EmTech China "Global Emerging Technology Summit", Kimberly Powell, NVIDIA's vice president in charge of medical and health, shared NVIDIA's development path in artificial intelligence.
"(NVIDIA) uses driverless driving to hone artificial intelligence technology, and then expands these technologies to other industries, including the medical field." Kimberly Powell said that Clara is an AI-driven car developed by NVIDIA A medical imaging supercomputing platform is used to improve the processing speed of traditional and old equipment for applications.
According to reports, the core of this platform is Clara AGX, which is based on the computing architecture of NVIDIA Xavier AI computing module and Turing GPU, and can be expanded from entry-level equipment to the most demanding 3D instruments. According to Kimberly Powell, the Clara platform can solve the problem of medical devices processing huge amounts of data of several gigabytes per second. She revealed that Clara has provided free use to early partners and plans to launch a beta version to specific targets in the second quarter of 2019.
This is just an attempt by NVIDIA in the field of AI medical care.
It is understood that as of November 2018, more than 50 medical institutions have invested in NVIDIA DGX series deep learning optimization servers and workstations, and more than 75 institutions have cooperated with NVIDIA to use AI technology in the medical field. , including medical centers, medical imaging companies, research institutions, start-up companies, etc. are its cooperation partners.
The following is an interview transcript of NVIDIA Vice President Kimberly Powell’s interview with All Weather Technology and other media, compiled:
Media: How has the Clara platform been implemented and accepted since its launch last year?
Kimberly Powell: Clara was launched in November 2018. We are also in the exploratory stage. Instead of opening it completely at once, we will first open online registration to interested partners. From the end of November last year to now, 350 to 400 companies have registered, and almost all the world's largest and famous corporate hospitals and start-ups have registered. However, it is still a very new thing, and there is no question of popularity and acceptance yet. The current version of Clara is the first version we have just released.
Media: What are the differences between the use of the Clara platform in China and other markets?
Kimberly Powell: The American customer is slightly more mature in IT, so he can execute Clara in the cloud. This is because the United States has data anonymization technology to implement Clara's cloud execution. The same set of software can run both locally in the hospital and in the cloud.
For the Chinese market, the support of a hybrid operating environment is very advantageous, because in remote provinces or rural areas in China, the network conditions may not be good and such cloud services cannot be obtained, so they can Choose to execute locally; but for those in big cities, they can choose to run in the cloud.
Media: What is Clara’s target user group?
Kimberly Powell: Clara mainly targets three major types of enterprise customers. The first is medical equipment companies, the second is artificial intelligence software development companies, and the third may be hospitals with hundreds of applications.
NVIDIA provides at least hundreds of different SDKs (software development kits) for developers in various fields. Clara is just one of these hundreds and is a toolbox for developers.
Media: What kind of operating model does Clara adopt?
Kimberly Powell: Clara’s development community is more about technical cooperation and less about commercial promotion. For example, Infer Technology uses the inference engine in Clara to execute multiple artificial intelligence algorithms in parallel. Without this inference engine, an AI model must be executed on a dedicated GPU. Therefore, for companies, Clara can implement their own applications in hospitals faster and more efficiently, using minimal hardware resources to run their artificial intelligence applications.
Media: How much does it cost to build such a platform in a hospital?
Kimberly Powell: Clara is not sold to hospitals as a separate software suite, but through NVIDIA's enterprise partners. Since it is installed as an application on the hardware system, it is difficult to answer how much Clara costs alone.
Nvidia’s hardware is a basic device found in almost all computing devices, so Clara is widely used. Even the game graphics card you buy can support the operation of Clara.
Clara is not only suitable for certain types of hospitals, but some hospitals may not be aware of Clara's advantages. They will gradually realize that no matter what computer hardware they buy, they can do three different types of calculations through the Clara platform, which is of great benefit to them.
Media: Are there any plans to improve the Clara platform in the future?
Kimberly Powell: Clara itself is a set of software, and the version currently released is still a relatively early version. Now we have some key areas to improve, such as interconnection with external hardware systems, support for communication protocols, and adding more acceleration engines to Clara to help start-ups accelerate the deployment of solutions.
At the same time, what we are doing is the transfer of learning knowledge and auxiliary functions. Hospitals in different regions have different conditions and use different equipment. We hope that the knowledge or conclusions analyzed on a certain hospital equipment can be popularized locally instead of just exporting the results. We should have the first such release by the end of January.
Media: What does NVIDIA want to gain from the Clara platform?
Kimberly Powell: The Clara platform uses NVIDIA's three important technologies, accelerated computing, artificial intelligence and visualization. In terms of medical imaging, we do not want three different workloads of computing, visualization and artificial intelligence to be executed on different hardware. We hope that one computer can do three different calculations through Clara.
For Clara, NVIDIA’s idea is software + hardware. In fact, Clara is also NVIDIA’s pavement for future smart devices. We believe that collecting data for post-analysis largely depends on what device you are using and when you collected the data.
In fact, we hope to empower medical equipment through software innovation. There will be intelligent equipment on the hardware side. At the same time, we configure the SDK for software development, which means realizing computing anytime and anywhere in the medical industry.
Media: What competitive advantages does NVIDIA have in the AI ??medical field?
Kimberly Powell: NVIDIA is more of an enabling company. Now many large companies are seizing the market for medical artificial intelligence. In fact, NVIDIA helps them better implement artificial intelligence medical applications at the infrastructure layer. , to help them achieve such market goals. Most computing devices use NVIDIA GPUs, and this is our positioning.
In addition, NVIDIA has a very large developer community. We have a CUDA SDK download kit with a monthly download volume of 500,000. Those who download CUDA are researchers from start-up companies or academia, so A huge foundation will also help those in the medical industry, because these will also be their customers.
Media: Who are Nvidia’s partners in the medical industry?
Kimberly Powell: We have four major types of partners, and this is true in every region. The first type of partners is in academia, because NVIDIA is not a doctor ourselves, we do not produce doctors, and we do not engage in medical research, so we need to seek partners in this area. We also have an NVIDIA Artificial Intelligence Laboratory (NVAIL), which is a global, formal partner program for this type of work.
The second type of partners is start-up companies. We have a project Inception (start-up acceleration program), where a local NVIDIA team in charge of the medical industry helps Chinese start-up companies. By starting this project, we can provide technical support to these start-ups, and the latest technology can be used by these start-ups first. If they have good solutions and products, we also help them with commercial promotion.
The third type is commercial partners in the industry, such as BGI and United Imaging Intelligence. We mainly cooperate in depth at the code development level, and we will also have joint commercial sales. support.
Media: What are the main aspects of cooperation between Nvidia and Chinese companies?
Kimberly Powell: We published a blog during the GTC CHINA conference. In terms of accelerating data science, we named the project RAPIDS, which actually represents the evolution of the NVIDIA platform: first starting with accelerated computing, and then going into depth. Learning, now is machine learning, and machine learning is what our RAPIDS platform represents.
RAPIDS is an NVIDIA platform targeted more broadly at the medical industry, not just medical imaging. After we released RAPIDS, we attracted many companies, including Ping An Insurance and BGI. Among them, Ping An Insurance has a large amount of claims data and medical data of insurance customers, and BGI has a large amount of genetic data. Even digital wearable device companies like Tanzhi Cube are very welcome to the RAPIDS platform. .
(For more exciting financial information, please download the Wall Street Insights App)