‘Artificial Intelligence has a huge potential to create new jobs in India’
BW CIO World met Jaap Zuiderveld, Vice President of Sales and Marketing at NVIDIA for Europe, Middle East, and India and Vishal Dupar, Managing Director of South Asia at NVIDIA to discuss NVIDIA’s foray in Artificial intelligence and its market strategy.
Edited excerpts from the interview:
BW: What is the go-to-market strategy for NVIDIA around the Artificial Intelligence offerings?
Jaap Zuiderveld: Today almost everyone from the governments to enterprises is talking about Artificial Intelligence (AI) and it was not the case about three years ago. It’s encouraging to know that there are early adopters and we are in the early stages of the AI. So, NVIDIA is on a mission to make AI pervasive, affordable & energy efficient. We are not talking about investments of millions of dollars. We have TitanX high performance graphic cards starting from roughly $1,200 that can bootstrap an AI computing platform. From here all the way up to $190,000, which is a supercomputer with VOLTA GPU’s that have the same power as 500 traditional servers in a data center. This affordability is what making AI pervasive and a savvy developer can start building AI models on a NVIDIA GTX.
BW: What are the opportunities for NVIDIA in India?
Jaap Zuiderveld: The Indian demographics are interesting. India has more than 50 per cent of its population below the age of 25 and more than 65 per cent below the age of 35. That is a huge demographic dividend, but that also means that India must create 20 million new jobs -- and I see AI as an opportunity to create jobs in India. How can we help the higher education community in India in their quest to understand deep learning and other AI technologies? By the way, we are looking forward to empowering enterprises. We work very closely with startups in India. There are around 160 -170 startups in India that are totally focused on AI and we know all of them. Most of them are 3-4 years old now and some of them have a huge potential to be companies of the future. We are starting with where it all begins: that is classroom and startups, as we are in very early stages of this market.
BW: When do you think will AI mature and become mainstream?
Jaap Zuiderveld: AI already has many applications today. The most common are Google and Facebook who are doing image recognition and speech recognition. So, it’s already touching our everyday lives. Of course, we see that North America is a little ahead of other markets but yes, we are sure that governments and enterprises from other regions will start working towards it. Interestingly, the prime minister of India, Narendra Modi, spoke about AI and mentioned that AI will dominate all aspects of human life very soon, and that India is creating an environment for uptake of various digital technologies through Digital India projects like smart cities and the India stack. Frankly, we were not talking about AI till three years ago, but the entire company has changed and is moving into AI solutions. The development on AI is going so fast and people are learning and picking up AI very fast. So, I think it will be mainstream sooner than later as developments and implementation will see an exponential rise.
BW: How will NVIDIA train Indian developers on AI?
Jaap Zuiderveld: NVIDIA wants to train about 100,000 developers in India in coming years via the train the trainer programs, inline with what we have been doing across the world. For instance, at SAP we trained about 25 developers who would take that knowledge and learning to a lot others in the SAP ecosystem. We are not going to make money on these trainings. We want to genuinely share our knowledge. For instance, IIT Guwahati has NVIDIA Tesla K20c GPU and NVIDIA GeForce GPU which they are using for breast cancer detection, and the HER2 scoring project. Similarly, access to these GPUs has been instrumental in training deep learning models at other IITs like Bombay, Madras and Roorkee.
Another example is the PARAM SHAVAK DL GPU, which is a supercomputer in a box for the academic intuitions and research organizations in India. It delivers a performance of 25PFLOPs on a NVIDIA Pascal architecture based processor (P5000/P6000). The National Supercomputing Mission will have a transformative impact on research quality and quantity by facilitating the training of Indian scientists and the development of “home-grown applications” in medicine, agriculture and technology. NSM will boost high-performance computing for India several folds.
BW: What are the kinds of jobs that AI will create? There is a perception that AI will lead to job loss?
Jaap Zuiderveld: When the Internet started, many people said jobs will be lost but in the end, it created more jobs. People have jobs today that we never thought of would exist before the Internet started. I think it’s the same with AI. AI will of course impact jobs and will create net new jobs. The Automotive sector has created many jobs around AI just because they want to know how AI will work in automotive. AI will create jobs in the BPO, KPO and high-end IT services sectors. An example of a BPO job would be fine tuning the AI models, performing quality control checks, customer service and exceptions management, dispute resolution etc. AI involves sensors, specialized components that will need to be designed and manufactured. The Make in India initiatives mean a lot of this will create new jobs in the manufacturing sector in India. So, there is displacement of products in many cases but not all cases are entire displacement of jobs.
Vishal Dupar: Let me add to that, AI will also improve existing jobs and let me give you an example. A radiologist can now have a second opinion from a AI system regarding the X-rays that he is investigating. And this brings down the chances of error as now the AI system provides validation and adds a layer of governance. There is a reskilling required in the India job market. We don’t deny that.
BW: Do you see a threat from Google TPU on the Google cloud platform? Oracle recently announced bare metal services and a data center in India. How does NVIDIA work with infrastructure cloud companies since some of them are also your customers?
Jaap Zuiderveld: First of all, Google TPU is only used in the Google cloud environment and it runs Tensorflow, which is an inferencing only product. You cannot train Tensorflow; it’s not a learning product. Compare this with NVIDIA and we support about 10 open source frameworks, while the Google TPU only has one, which is Tensorflow.
SAP will soon start working on GPU accelerated cloud systems and we have many projects underway; I cannot share that information right now. We work with leading cloud providers across the world. Baidu, AWS, Tencent are cloud providers that have GPU acceleration options.
BW: Enterprise databases still don’t support GPU acceleration. What are your thoughts on that?
Jaap Zuiderveld: We are already talking with Oracle and SAP like I said. Look at our NVIDIA DGX-1 which is the world’s first supercomputer for AI-accelerated analytics. DGX-1 applications partners, solutions, technology, service partners and research project partners are continuing to improve and grow. And DGX-1 comes with GPU accelerated in-memory databases like Blazegrah, BlazingDB, Kinetica and MapD that are enabling some fantastic use cases for the NVIDIA DGX-1 platform.
BW: Enterprises are looking at both AR/VR and AI/machine learning. Which among them will move faster in terms of use cases and drive greater adoption of GPU’s in the next few years?
AR/VR is relatively new compared to AI and machine learning. The research labs have been investing in the science of AI from many decades now and thus AI/machine learning is something more realizable in the short term. AR/VR is more consumer focused and I think it’s a little early for that becoming mainstream. But we are investing in both as eventually the base architectures of our GPU for both the applications are common. So any investments in the GPU; by the way NVIDIA R&D budget is 25 per cent of total revenues. All this spend goes into taking GPU architectures to the next level and to bring incremental value to the market on a consistent basis. The base platform is the same and we are working on improving it every day. The same platform can enable both AR/VR use cases as well as AI/machine learning.
_20240910143022_original_image_4.webp)
