Firmly disagree that AI needs to be regulated: VMware’s Pat Gelsinger
Pat Gelsinger has been chief executive officer of VMware Inc.—a unit of EMC Corp., which is now part of Dell Technologies—for the last five years. EMC was acquired by Dell last September but VMware continues to be run as an independent unit. Gelsinger, who earlier worked with Intel Inc. for 30 years and even became its first chief technology officer (CTO), holds seven patents in the areas of VLSI (very large-scale integration) design, computer architecture and communications.
Gelsinger has been visiting India at least once a year for the past 25 years. In an interview on the sidelines of VMware vForum 2017 held in Mumbai on 23 October, he shared, among other things, his views on digital disruption and newer technologies like artificial intelligence (AI). Edited excerpts:
Now that you are part of Dell Technologies, how is the synergy working out with Dell products?
It is exceeding our expectations. We built a business plan before the merger was announced, with regard to the detailed models, number of servers sold, number of regions and market share. I have already announced that we have increased the synergy targets for Dell and VMware. There are places like China where Dell has over $6 billion of business.
How big is the market for VMware in India?
We have had great success in the Indian market, with the market being our largest site. We have every function at VMware represented in India, as well as the R&D (research and development) vertical.
You say all businesses are digital businesses—that there is no longer a distinction between a traditional and a digital business. Give us some examples of this trend.
Consider General Electric (GE). They are into engines, oil rigs, power plants and are now focused on software skills because they need analytics to adjust their turbines. GE is also building smart power grids that both store and deliver energy, which accounts for up to 3-4% of the improvement that makes an extraordinary difference to efficiencies.
Another example is Bosch, which was once an old-line industry but is now becoming connected and Internet of Things (IoT)-centric, and needs to be managed and automated. Honeywell works on HVAC (heating, ventilation and air conditioning) systems that are now being connected, automated and secured. Retailers today are innovating for new shopping experiences, to compete with players like Amazon. Hence, their online store experience becomes a digital experience.
Some of the big disruptions are Uber and Airbnb. One is the largest transportation company but owns no cars. The other is the largest hotel chain that owns no hotel rooms. Every industry is being revolutionized by technology in some manner.
But even as we talk about digital, we refer to the phygital (blending the physical and digital) model. Do you see this as a contradiction or a logical conclusion?
I think in many cases they started out as contradictions and almost thought about their business in two worlds. On the one hand, you have these new technologies you want to work on. On the other hand, you may want to build on the old technologies. It is not like the power engine in-charge at GE is no longer important, but now we are bringing digitization, new analytics, machine learning and intelligence into that aspect. Everything needs to be innovative if you are truly going to transform the business. And to do that, you have to integrate those two worlds together.
Speaking of new technologies, cloud computing companies are integrating their services with AI to differentiate themselves.
The field of AI is definitely exciting and has been around for 30 years. Almost all of the core innovations in AI happened in 1980 and in the early 80s. What we are seeing happen over again is that these technologies need about 30 years to mature. That is exactly what we have seen happen with AI. What we are also seeing in the industry right now is that cloud, Big Data and IoT are being applied at a scale unlike ever before, owing to technologies like machine learning and image recognition. Today you can put millions of photographs together and be able to run the algorithms on those millions of photographs to identify whether it is a human or an animal. Cloud will be an important aspect, but we also think that it needs to be applied not just in cloud, but also in self-driving smart cars.
The sense I get is that you do not fear AI and are rather comfortable with the kind of progress that AI is making.
Yes, absolutely. I firmly disagree that AI needs to be regulated and in that sense, I believe that it becomes an augmentation. The fact that a doctor can now have an AI assistant to look at a cancerous cell image, augmentation is definitely a powerful tool. Imagine the world we would have today without the spreadsheet. I think of AI the same way. Why wouldn’t one use this powerful tool to its best advantage, one that doesn’t replace but instead augments? I am thrilled by the possibility of being able to take several simplistic things and do them at a scale and speed unlike ever before. As we go forward, I think we will find AI being increasingly integrated into products everywhere.
Having also served as the CTO of Intel, what are your thoughts on quantum computing? You do not seem to think that it will replace general purpose silicon-based computing.
Quantum computing is very exciting in terms of a research area and lots of people are starting to show some more progress, but it will never be a mass-market technology. Quantum computing is exciting because there are a few problems that it can tackle uniquely, better than anything else. Some of the security problems are those where we are able to have many parallel quantum states that allow you to probabilistically look at certain problems in ways that were never possible before.
So how do envision the future of computing, given that Moore’s Law has its limits?
Moore’s Law continues but it is moderating. It used to double every two years—now it doubles every three years as we’re getting close to the fundamental limits, but computing hasn’t slowed down. What is increasingly happening is that we have applied more computers in a distributed fashion, rather than having one computer that is twice as fast. That is why AI is interesting, because I can now have a cloud—so instead of having one big computer or supercomputer, I now have everybody who can rent a supercomputer, which is essentially thousands of little computers tied together.
How is your strategy, comprising private cloud, the software-defined data centre, hybrid cloud and unified communications, working for you?
The first one is foundational—moving from virtualization and computing, to the full data centre. That is what you call the software-defined data centre. We launched our virtual networking product with NSX, our virtual storage product with vSAN and our automation with vRealize. Our automation ranks number one. NSX is clearly seen as the market leader today. For virtual SAN, we just announced 150%-plus growth rates in the last quarter. Compared to where we were five years ago when I took over (as CEO of VMware), we did not have any of the other products.
For end users, our Workspace ONE offering is doing extremely well because we had brought together the virtual desktop, AirWatch mobility, identity and common management. We can now go to customers with the mindset that it does not matter what kind of device you have or work with—be it HP, Samsung, Apple, Dell or Microsoft—we can manage all of them in the same manner and provide you with the best user experience across all those devices. Today, customers increasingly want fewer vendors to do more. They do not want to integrate all the pieces together, and instead want to get a larger solution from their key partners. That is something we are now able to do.
But security remains a big challenge…
If I think of the technology industry as a whole, security has failed customers in this area. Over the last five years, vendors have spent more on security than any other technology, and yet there have been several costly breaches. Customers have spent more and they’ve fallen further behind, which is not good. The technology industry needs to build security directly into our products and not ask customers to turn things on. So, we have started to announce products and capabilities in this area specifically, and I think over the next year you’ll see us step up in a big way.
Partnerships are clearly one of the foundations of your growth—for example, your tie-ups with companies like Amazon Web Services, Google, International Business Machines Corp. (IBM) and Fujitsu. What’s the way ahead?
The idea of partnering to deliver a more complete solution is very important for VMware. Our technologies are often part of the bigger solution. For example, with Fujitsu, they are becoming a big partner for delivering IoT. They are focused on auto companies like Toyota, as well as industrial companies where they are strong—for factory automation and retail. We also announced Google as a key partner for containers (ensures that applications run quickly and reliably regardless of where they are deployed).
This idea of containers as a new way of developing and deploying applications is gaining a lot of momentum in the marketplace. Kubernetes is a key technology that Google has created. We are taking our technologies and combining them with Kubernetes and Pivotal (a Dell Technologies unit). We believe this will be an important way for us to reach the market.
IBM is taking VMware and adding its technologies like security solutions and Watson analytics, and taking it straight to the marketplace. We just announced a major customer win with IBM at Lloyds Bank in London.
In terms of the hybrid cloud or multi-cloud, we have the IBM partnership and the vCloud Air Network partners but the Amazon partnership was a game changer because the number one private cloud company was coming together with the number one public cloud company to deliver a seamless customer experience to move from private to public or public to private.
Speaking about new technologies, your thoughts on 5G networks.
We are very excited about it. 5G will be the largest capital investment cycle for the rest of our careers, because it will be a new spectrum. It will be new to antenna, base station, radio and networks. It will also be new to how people operate their core networks and that is where technologies like NFV become very important. NFV definitely excites me the most because we are moving on from virtualizing the data centre to virtualizing the network. (Network functions virtualization, or NFV, is expected to reduce the amount of proprietary hardware needed to launch and operate network services.)
When you talk about 5G, I think about it as a 100x and a 100x. On one dimension, you want bandwidth that is a 100 times greater and on the other end, you are going to have a 100 times more things connected to it with IoT—right from smart devices to smart cars to smart cities to connected machines and analytics.
Today we have about half a billion people and about eight billion smart devices. By 2025, it is expected that we will have 30 billion connected devices on those networks. So, it’s going to be the explosion of machine-connected devices that we are going to see. Hence, this presents a major new opportunity as well as a challenge for building and operating in those networks.