Cows and data centers: What HP’s chief engineer thinks about fusing the real world with technology

Chandrakant Patel has a deep history working on hardware and fundamental science at Hewlett-Packard, and he has used that background to create a vision for the future of technology that combines the physical and digital worlds.

He hopes to inspire his fellow HP colleagues and the rest of the tech world on a new decades-long path. Patel is a senior fellow and the chief engineer at Hewlett-Packard. That’s an important and rare position, as HP has more than 50,000 employees in 170 countries, with many thousands of engineers. I met Patel at the 50th anniversary of HP Labs in Palo Alto, and we caught up for an interview after that event at a very special place for HP employees: the original garage at a home on Addison Street in Palo Alto, Calif., where HP was born in 1939.

Patel’s job is to inspire HP’s engineers to be creative when thinking about the big technology problems they must overcome. After all, Moore’s Law — or doubling the number of transistors on a chip every couple of years — doesn’t just happen. It is the result of a lot of smart people figuring out the toughest technical problems of the day. Patel believes that we still have to figure out a much more energy efficient world network, with intelligent devices at the edge that don’t drain resources out of the data centers.

We talked about why he chose to stay with the PC and printer maker, HP, rather than HP Enterprise, the services company, after last year’s split-up. And we reminisced about HP’s past, including the creation of its first computer 50 years ago this week. Patel is very passionate about how students should study the fundamentals of science — and both hardware and software — to prepare themselves for the age of the Internet of Things. He prefers to call this the “cyber physical” applications, which expose the seams between hardware and software, between the real world and the digital.

We chatted there so that we could get inspired about the history of technology and where it’s going in the future. Here’s an edited transcript of our conversation. I’ve also added many of Patel’s slides, as he loves to paint his ideas of the future by making sketches.

Chandrakant Patel: I’m a mechanical engineer. I started at an interesting time in Silicon Valley. My first interview was with a company called Dysan. It was on Patrick Henry Drive. Patrick Henry was brand new. Now the stadium is very close to it. They were making disks. Heads and media were done here. I got a job at Memorex, where Nvidia is now located.

A long time ago, Memorex had that commercial – “Is it live or is it Memorex?” Ella Fitzgerald would shatter a glass with the frequency of her voice. They’d copy it to a Memorex tape, and then playing back the tape would shatter the glass too.

The reason it’s important to me is it was a prime time commercial. People understood why the glass broke. People understood physical fundamentals, back in the early ‘80s. I found myself in what I called the “valley of tinkerers.” Memorex had its share. Al Shugart, Finis Conner. They went on to create Seagate. We had manufacturing and design there.

I was making drives where the mass was 100 kilograms. A gigabyte would cost $100,000 and it was the size of a washing machine. Because the mass was very high and the stiffness was low, the frequency, the characteristic frequency of the drive was low. Low-frequency vibrations could damage it. As mechanical engineers we had interesting problems to solve.

VentureBeat: It was an age of physical hardware.

Patel: Very much so. Understanding how physical hardware worked. Discs were rotating at 3600 RPM, 32 heads, how do you keep them flying? Then one thing I noticed was, as the hardware got smaller, the drives got smaller, I felt the stiffness was going up. The mass went down. The ratio of stiffness to mass goes [up] and the natural frequency goes up. They’re less susceptible to those low-frequency vibrations. It was simple first-order fundamentals-based thinking to see that drives would be commoditized.

I reset myself, after working on large drives and small drives. I joined HP Labs in 1991 to work on the PA-RISC chip. We were going from wire bonds to the flip chip to get a lot more I/Os out of the chip. I established the chip packing and thermal management. I did a lot of work on electronics cooling. That’s when I got to know Bill. Subsequently I felt that chips would come from one or two places. As you scale down you need volume.

I moved out into systems, working on large-scale systems like Superdome, the supercomputer-class systems we were building. In the mid-’90s I went to my boss and said, “The data center is the computer. The building is the computer.” I filled a room with racks that I said would be about 10 kilowatts, filled with industry-standard components. Now the building is the value add, not the servers. It is the networking, the cooling, the power—power, ping, and pipe. Those three Ps would determine the data center and the total cost of ownership of a data center is driven by energy.

My boss said, “Why do you want to work on facilities?” My contention was, it’s Carnegie Hall with 150 people per seat. A person is 100 watts. A rack would be 15 kilowatts. That’s 150 people in a seat. Imagine that. You have to deal with fluid flow and so on. We created the smart data center project. We build a data center with sensors and controls. We built the dynamic control systems for it. We were the first ones to do that.

We went on to build Eco pods. That started because of a conversation with a customer of ours. The customer had underground mines. My recommendation was to put data centers in containers and lower them into the ground. The region where they were, the ground was nine degrees Celsius. I said, “Let’s dump heat into the ground.” That didn’t happen. I wish it had, but the dot-com boom and bust happened at that moment. Otherwise that would have been one of the most secure places in the world.

VB: I understand that’s a big business in Iceland now.

Patel: Right. The motivation for Iceland was low-cost geothermal electricity, cooling, the external layer. But I looked at three things. Not just cooling, but power, and also ping, the networking element. I felt that the data center costs needed to be reduced significantly to connect the entire world.

We worked on what I call the first and second generation. Then I went on to drive the third-generation design, where the data center should be completely off the grid, for example run with micro breeder power supplies. It could be solar. But one of my favorites was that data centers should be integrated with dairy farms. A big commercial dairy farm in the U.S. might have 50,000 cows, and only 2,000 cows could generate a megawatt of electricity. The anaerobic digestion of manure generates methane and you use that to run a standard engine. G.E. makes one. It’s a very good engine. Use that to generate electricity. It’s systemic innovation. You just put things together.

Then we considered that the heat from the IT equipment is low grade. Generally we just dump it. Why not take the hot water and put it through the manure in those big digesters? Farmers call it soup. If you dump the heat into the soup, you enhance methane generation. Then use methane to generate electricity. Even the waste heat from the generator is high grade. You could use that to drive the absorption/refrigeration cycle. Our contention was that IT and manure have a symbiotic relationship. [laughs]

VB: Cows and data centers.

Patel: Right! At the time we wrote a paper. It didn’t get much interest. But when we said that IT and manure were symbiotic, it took off. We got a lot of press. Farmers invited us out. So we wrote another paper and contended that when you build a data center, you must think of your own power station. You must make sure your demand side matches the supply side. All work loads are not equal. If someone is going to pay more to execute his work load right away, execute his job before the guy who can wait. We built a whole software around dynamically provisioning the demand side balance. I felt that design was needed. The industry doesn’t seem to have gone forward with that.

Data centers, to a large extent, have not succeeded. 60 percent of the world is not connected. There’s talk of using satellites and drones, but I don’t think that will work. I felt it was time to move again. When the split happened and I had a choice between HPE and HPI, I said, “You know, data centers will remain. I know it. We’ll have a hybrid opportunity. But I should go back to the edge.” The edge devices, in a peer-to-peer network, will connect the rest of the world. It will happen south of the internet. Maybe even at cell towers.

The reason why I think connectivity will happen at cell towers, why we’re going back to computing at the source, is twofold. One, people are not connected. In places where they’re not connected, though, telco is prevalent. When I go to India, in my wife’s village where the leopards still roam, I get a better connection than Hanover Page Mill. They have great voice. Cell towers are everywhere. But only 250 to 300 million people have data services. A billion people don’t. And yet those are the people who need it most.

When I go to India I get a train reservation because I’m connected. The guy who makes five dollars a day has to go stand in line for the same seat that I take. It’s a digital divide. The prime minister wants to connect them all. My contention that a hybrid model will evolve. There will be data centers, but the action is going to be at the edge.

One of my goals at HP is, what does edge computing look like? The other thing on the edge I think about is, what are the cyber physical applications? We have failed, in many cases, to address those. In the 21th century, those of us who were connected, we followed tweets from Ashton Kutcher or Beyonce, but we failed to follow tweets from airplanes. We lost airplanes. We couldn’t follow them. And yet an airplane generates a terabyte of data. We should be analyzing it, tracking it, helping the pilot. When the pilot fell from 35,000 feet because his Pitot tube was frozen, we could have helped him.

To me, a whole class of physical application hasn’t been addressed. The 21st century is about the integration of cyber and physical. If I want to analyze things on physical systems like airplanes, I’m going to have to do analysis on the airplane and provide real-time insights.

Events might go back to the data center. We may collect data for six months to do historical analysis of the entire fleet. A hybrid model will evolve. To me, computing at the edge is very important.

VB: So there’s an awful lot of computing that happens in a local space, but you don’t always need to go back to the main internet?

Patel: No. Robots today traverse pipes. They take video data. That video is sent to Bangalore and people sit in a room looking for cracks in the pipe. Shouldn’t we be more sophisticated than that? I contend that tomorrow, the robots will go in the pipe like they do today. We have an aging infrastructure. Water mains break in the bay area all the time. That inspection will happen, but the video data – which is terabyte scale – will have to be analyzed on the robot itself. Maybe we’ll also close the loop with autonomous systems where the robot traverses the pipe and fixes it as well.

That’s the world we’re getting into. If we dream big with autonomous systems, there will be a lot of local computing and local action. Then there will be data centers that collect the information for historical data mining purposes. That’s one big driver.

VB: Back to the peer-to-peer model, it seems like you’d put a lot more computing into the edge to make up for the lack of bandwidth there. I used to think of that network computer model that Larry Ellison would talk about. Put a small computer here, connect to the network, and the network provides the computing. As long as the connection is good, you don’t lose anything. But if you’re talking about the edge, where you don’t have the bandwidth, does what you put in the computing on the edge matter a lot more?

Patel: It does. The trend I’m seeing—I’m a mechanical engineer. I did CAD, CAE, CFD. Better and better workstations were built to do that, and then we had a general-purpose computer that worked. I see in verticals – transportation, biomedical–if they do lensless microscopy for imaging blood samples or something, they’re hacking together computers. But that doesn’t scale. We must think of ways we can scale.

As a computing company we have to ask what the architecture looks like. Maybe it’s a performance machine related to just a pig in a pipe, or in an airplane. But what does it look like? Even though we’re addressing a vertical, how do I make it flexible across many verticals? Is there an FPGA?

When we did dynamic smart cooling, we ran the control system on Matlab and Excel. We would do the calculations, go to Excel, change a cell, and then the cell would change the blower speed. The whole data center ran like that. If I can do Matlab code and say, “I’m going to put it on that computer in the field,” can I have an FPGA so I can use my code in a world with more and more Teslas and the like where you’re updating physical systems? Can we go beyond updating firmware to creating a completely new logic?

It’s high time that Silicon Valley came back to the cyber physical world. This hurts me, because when things blow up, whether it’s in a factory or—the pressure was building for days. If only we followed that the way we follow the frivolous stuff that happens in our world. It’s what I call operating at the crossroads of people, profit, and planet.

VB: Do you feel this fits within the whole Internet of Things theme?

Patel: I use the terminology “cyber physical systems.” It does fit into the internet of things. The only thing is, it’s not clear to me that you have to be connected to the internet all the time. Surely, for security purposes, you may choose to be sometimes connected to the internet.

VB: Smart and sometimes connected.

Patel: Right. That might be the right terminology. Smart to do what you’re doing locally. When you need to use the internet, you use it. But you’re not full time. So it is the internet of things, but it could also be just a physical and a cyber system.

In a lot of places we won’t get the connectivity. If we’ve failed to connect 60 percent of the world, what makes us think we’ll connect these trillion devices?

VB: It seems like there are still a lot of problems all over the place. If the internet of things happens, then you get billions of devices connecting to overloaded data centers. You have to work on data centers again.

Patel: I just did a back-of-the-envelope number. Let’s say the U.S. has 20,000 data centers, each about a megawatt, so the U.S. needs 20 gigawatts of power for data centers. That’s about 20 nuclear power plants. Now let’s say that the U.S. model is replicated in India, to connect 1.3 billion people. If I scale based on population alone, I need 80 gigawatts. India’s total energy production is roughly 300 gigawatts. How can we have data centers take up almost a third of the country’s electricity production?

You have always-on with a lot of spinning systems, even with solid-state drives—I ask myself, are we more efficient if we control our own destiny? Do we switch off the lights when we own the lights? If I have a local computer, will I manage the power better locally? We have to ask ourselves if we’re going back to the ‘80s. [laughs]

If you look at the systems we’re building, like 3D printing systems—the digital pipeline is very interesting. We can create a voxel that’s 20 by 20 by 80 microns. To make a part, you have trillions of voxels in there. If I have the ability to modify each one of them someday and create new types of materials, the digital pipeline is massive. What kind of computer do I need for that?

HP is in a unique place. We’re a physical company and a cyber company. You just saw Bill’s garage. When I joined, I went to Corvallis, to Vancouver, to Barcelona. It reminds me of the HP I knew 30 years ago. People with the same fundamentals, who care about pool boiling, nucleation—somebody might make a cup of team in the morning and just walk away from the pot of boiling water. We look at the bubbles forming, nucleation happening, bubbles coalescing, bubbles departing. We go back to fundamentals, like boiling heat transfer. That’s how the inkjet came about.

Even today I find that a lot of the physical scientists in our area, in the labs, go back to fundamentals. For nano-scale structures, how do I enhance Raman spectroscopy? They’re looking at Raman response, going back to fundamentals. This is a strength we have.

VB: It’s a good match for the maker community, it seems like, all this renewed interest in hardware.

Patel: Right, I would say so. As we put up our own maker spaces — I like to call them “tinker tanks,” given my own connection with Bill – as a chief engineer I want to make sure we have a pipeline of problem statements for them. We create the environment so you can go and play, but I want to make sure—the problem statement could be anything. It could be as complex as modeling heat transfer in 3D printing. It could be systemic, like how to use my devices to do physiological monitoring of a human being.

Things that are done by sophisticated devices, can I just use my cell phone to do all the critical physiological parameters? Today, more and more accessories are available. How do I put them together to build a systemic innovation? Just like that earlier example I gave with the data center. There is systemic innovation, and then there could be innovation that’s very fundamental. We’re trying to create a culture like that in the maker space, a very cyber physical culture.

VB: A week ago we had the hacking of all the IOT devices causing a lot of problems. Security still doesn’t seem like a solved part of that situation. It doesn’t seem like you can go forward with it until that’s addressed.

Patel: I can share a visual I use. I also drive security. There’s a research team and then an HP-wide security initiative that I drive. The way I see it, the 19th century is the machine age. The 20th century is the information age. The 21st century is the integration of those two. How do I use haptic holography and information analytics to manage my system? Very much value-driven.

When we think in terms of security, whether it’s collecting data here, like the video in a robot—I have to think about security from end to end. We hear about cybersecurity, but what is cyber physical security? At the hardware level, what kind of security features are we going to have? Am I connected all the time? Why am I connected? We have to ask all of those important questions. If we’re having this issue today, it’ll be a nightmare in a world where all these physical things are connected.

VB: Charlie Miller, the hacker, gave a talk at the ARM TechCon event last week. He said, yeah, everything is all connected on the one CAN bus in a car. Could you have done two different systems? Maybe one for infotainment and all the systems for driving on the other? They didn’t do that. At the very beginning, they didn’t think anybody would hack a car.

Patel: Exactly. And what about an airplane? How do we segregate the entertainment system from the critical systems on an airplane? The Iranian nuclear centrifuges, they weren’t connected, but somebody used a USB key to get into their building management system.

For the kinds of things where I need real-time action, where I’m doing integrated supply and demand management, I need to be local. This is how I divide it. I go north of the internet when I’m doing historical analysis. I worked on analyzing cooling towers. We were building data centers in India and there were disputes over water, so we didn’t want to waste any.

We collected data from these cooling systems over a six-month period and looked at the patterns as they operated from a thermodynamic efficiency point of view. We chose the pattern that gave us the most efficiency in electricity and water consumption. This is an example where data was collected. We determined the pattern. A subject matter expert sits down with a machine learning expert and determines the best pattern to use.

The other thing I feel that is happening is the rise of multidisciplinary innovation. It’s no longer unidisciplinary. You can’t sit above the internet abstraction layer and create apps without knowing how a compressor works. Or you can, but you could do a lot of damage. In order to drive something useful—you can look at a compressor and say it’s cycling so many times and so it’s short-cycling. But a mechanical engineer might say, if it short-cycles X number of times it’ll fail. This is an area where I think Silicon Valley needs to be prepared not only with people with cyber skills, but physical skills as well.

I have a big concern there. I go back to my Berkeley days. It was all about fundamentals. Cory Hall had the microelectronics lab. I wonder how much focus on physical fundamentals is there today.

VB: I wonder if there’s a difference between the people who are working at HP versus people who are working at, say, Facebook.

Patel: Could be. Or other social networking companies. I write a lot on LinkedIn about the dawn of the cyber physical age. I had the fortune to be a mechanical engineer, but also gain that breadth in software design and system architecture. One control system we built had more than 10,000 sensors, more than 600 actuators. It was implanted using .NET. So I didn’t gain coding skills, but at least architectural skills. When I have these conversations, sometimes I do hear people saying I’m from a bygone era.

One of my neighbors is a civil engineer. He likes to say that the guy who built the Golden Gate Bridge never made millions. But 50 years from now I’ll still be able to drive to Marin County. One guy told me that there’s nothing new in civil engineering, so I sent him a video of automated bridge-building in China. There are new things happening. We just don’t want to explore it. Civil engineering is just as important.

I’m looking at the idea of being able to have tactile interfaces. We’re concerned about people who run power plants who are retiring. What about the mechanics who maintain airplanes? As an IT company, as a cyber physical company, how can I look at AR and VR and haptic holography to improve productivity? One person might manage a lot of different airplanes. Will haptic interfaces help? When I look at all these cool things, I have to look at it in terms of productivity improvement, energy savings, and value to the customer. Will the customer buy it from me because they see value in it? Everything is a value-driven conversation.

My data centers were the same. I remember when I talked about energy savings in data centers. I went to a conference and someone told me, “Nobody ever got fired for wasting energy. You get fired if a server shuts down.” My reply was, “You’ll get fired for both.” If your service agreement says the server should be on, it should be on. And if you’re saying energy doesn’t matter, when I went to the customer, they’d say, “If you give me a six-month payback, I’ll buy it.” This was in 1999. Customers are always looking for value.

I think of myself as the chief engineer who’s on the bridge looking at trends and thinking about what technologies we should worry about. But I’m also in the engine room. We have a very multidisciplinary engine room, because of our breadth and our diversity of products. So how do I surface those skills? We’ve gone across the company and created a unique model. We’ve surfaced these skills in materials, micro-technologies, system-on-chip, thermal management, liability, all the way from bottom to top. These are the slices that are necessary to build value stacks.

VB: Hardware and software?

Patel: Hardware, software, firmware. There’s software, data science, knowledge discovery. We have our firmware group. I believe the cyber physical world is a hardware and software co-designed world. We can’t be over-provisioned on software and under-provisioned on hardware.

In our case we’ve identified these communities. They come together and teach each other and drive guidance. My vision going forward—let’s say a business unit leader says they need help with figuring out how to test things at high temperatures. There aren’t industry standard tests available. I should be able to mobilize the materials team, the liability team, the thermal management team, and create a special ops team on a helicopter that goes out and helps that leader. It’s on-demand provisioning of resources for whatever value we’re delivering in a vertical – robotics, security.

To go back to security, I think of security as a horizontal and as a vertical. If I have a cyber physical system, how should I think about security from bottom to top? How should I think of security horizontally in every device that I build?

VB: I had fun talking with Shane about science fiction and video games. They’re often very fictional, but some of the games coming have a way of simplifying to get a point across. Watch Dogs 2, from Ubisoft, comes out in November. It’s set in San Francisco and it’s all about the future when we have a smart city. Hacktivists fight back and they hack into the smart city, for a cause.

The developers were saying that the interesting thing about Silicon Valley is that a lot of technologists here in the Bay Area are thinking not about tech for tech’s sake, or tech for getting rich, but keeping technology open. I wonder how much that idea is interesting to you.

Patel: You can look at an example like the blockchain, Bitcoin, the existence of a distributed ledger and the whole motivation there, so there’s no central authority keeping track of transactions. We pay attention to that. What I ask—it’s not necessarily for digital currency alone. What does it tell us about how we should think about models evolving in the future? As these models evolve, how should we think about computing? If distributed ledger blockchains are an example of the way things are going to be done, given our core in computing and printing, what should we think about as far as how we design these things?

My philosophy, as I say, goes people, profit, planet, and petabytes of data. How do you operate at that crossroads? On the city scale, it’s the same thing. How do I do city-scale resource management? There was a time when we motivated cities and thought about city-scale architectures.

The interesting thing is, a lot of cities aren’t able to do the investments that are necessary. Many cities we talk to want to do resource management as a service. How do we help cities on a pay-per-use basis? As you think about these models, are there ways we can help cities manage without having to put out the necessary capital outlay?

VB: They get to a very particular contention point. Can you build the internet of things and smart cities and still have some shred of privacy?

Patel: When we’re looking at AR and VR, going back to haptics, I saw something with a workstation that was showing the insides of a human body – blood vessels and so on. My son, when he was studying surgery, I asked him, “There’s so much in there. How do you know what you’re touching?” He says, “You feel the vessel for elasticity.”

When I see a lot of these games—I try to put together things that people have to do. Games, AR, VR, they can provide applications in various verticals, whether it’s simulating surgery or whatever else. That’s where a lot of haptics things came in for us. Haptics is a great dimension in that regard. If that’s how they’re already training surgeons, how can we provide that feeling in a simulation as well?

VB: Going back to some of your interest in the edge, we have data centers. We have smartphones. Do you feel like there’s something missing from this basic equation? A new area of computing that has to emerge to fulfill what you want?

Patel: One image that I think about, I go back to our laser cartridge days, formed cartridges. I may be biased, because from a sustainability perspective I think in terms of life cycles, cradle to cradle. I don’t want proliferation of devices. I want to make one device work for a long time. And I look at the computer my dad had, an all-in-one back in India. He bought that because he wanted to put Magic Jack in it to make free phone calls on his 56 Kbps connection. He didn’t want to make the investment unless he could get a payback.

We’re building these computers. Can we use them to help? We have an aging demography. More people are over age 65 than under five. I have computers with fancy cameras and sensors. How can I read emotions? How can I use machine learning to do more inside the home? When I start asking myself that, I feel like there’s something missing between mobile devices and workstations.

There are people driving these situational intelligence algorithms. They get a lot of data – six months of data, 12 months of data – and they’re oblivious to the hardware at the bottom, but they’re coming up with algorithms, advanced algorithms. If they come up with an advanced algorithm, how do I impedance match with it? If I sell you a piece of hardware, and I come up with an advanced algorithm a year down the road, do you have to buy another computer? Could I have sold you a computer where now you can download the latest algorithm? I’d like to do the latter. I’d like to sell you one very useful computer, just like that Magic Jack was for my dad, except now you’ll download that algorithm and all of a sudden you get a new set of value.

I’m inspired by our neighborhood friends here, the flexible models, like the Tesla guys. They sell hardware with all kinds of flexibility. How do I build a platform like a car? You don’t just go out and buy another car. If you’re going to make it autonomous you’ll have to put physical features inside it. Similarly, how do I build a flexible, configurable, upgradable computer system? What’s lacking is reference architecture for a flexible, configurable computing building block.

My challenge is, what about the FPGA? That’s what I go back to. People say that FPGA programming is very complex, but I look at research at UCLA and other places where they’re looking at building an abstraction layer. You could conceivably someday take Matlab code and put it on an FPGA. The software guys are advancing. Algorithm guys are advancing. How do I do hardware so I can keep up with them? That’s where the software and hardware co-design comes in, and that’s where I think we have a missing link.

VB: Is this revisiting some of the failed efforts to do things like software-defined radio and reconfigurable hardware? Hardware that can redo itself.

Patel: The idea of making something self-healing, resilient—I don’t know if that’s it, per se. But it could be in the sense that you have something that’s low-cost and easy to build with it. I feel like the reference architecture is lacking, again. It’s more about how I enable things where you don’t have to pay a lot of money, but it’s sitting dark until it’s necessary. You don’t have to upgrade.

In a mobile device today, are we curtailing performance because of heat? If I have applications like at home, how do I drive a Lamborghini on the autobahn and not frequency scale it? Sometimes I wonder why we’re leaving all the performance on the table. I feel like we’re driving a Lamborghini on city streets, as opposed to the autobahn.

The thermal management problem can be solved once you dock it. I use the Elite X3. When I dock it, can it go into a super high performance mode? Can I use that one computer? Can my dad use the Elite X3 in a high performance mode when new features come in and require more power, but it’s in a docked mode? How do we advance that?

VB: Nintendo had some of this idea with the Nintendo Switch, where they’re going to have a sort of portable tablet that plugs into a dock, which you then turn into the home console.

Patel: That’s what we did with the Elite X3. You plug this in and it’s Windows 10. I can use Powerpoint. What’s the next generation of Elites that we’ll have? Maybe we’re paving the way for that, for useful applications.

VB: It sounds like you’re open to maybe reviving some ideas of the past that are now doable, or more practical.

Patel: Sure. Leverage the past. The performance we left behind because we felt there was a thermal management barrier. How do we address that? As someone used to thermal management—we came up with all kinds of cooling systems. The one we thought we needed for a very high-performance chip was to spray coolant with inkjet heads. You provide the right amount of coolant so it matches the heat flux. If you use too much, a bubble forms. If you use too little, a dry-out happens. But you change phase all the time. We demonstrated the highest heat removal. But we never got to that, because instead of trying to solve the thermal problem, we decided to curtail performance. When multi-core came along, we decided to use that.

But what if I want to run all the engines at full speed when it’s docked, so I can do useful things that my friends inventing these algorithms can leverage? There’s an impedance mismatch. Suddenly that mismatch will get bigger as we go down the road.

The other thing that drives me is 3D printing, particularly because I’m a strong believer in cradle to cradle and lifetime joules as the currency, not dollars or rupees. I met Alan Greenspan once. My contention was, if joules were our currency, would we be digging bauxite out of Amazon jungles to ship to China and make aluminum that we ultimately threw away somewhere else? Or would manufacturing and design be local?

We’re getting to a world where energy is the currency. Cell phone users—almost 80 percent of the elements from the periodic table are in a smartphone. We’re using up critical elements, which will lead to geopolitical unrest. We need to think of ways where we can reduce this resource consumption and think from a holistic perspective.

That’s why I’m very excited about 3D printing. If we can make materials—it’s not just about making parts. There’s all this excitement about making parts. 3D printing parts is great. But can I also make materials on a voxel by voxel basis, mixing things? In the old days, if you wanted to use a composite—say I want to use aluminum silicon carbide to match the coefficient of expansion in thermal conductivity. I’d go to Alcoa or some company and get aluminum silicon carbide slabs. I’m using a material that’s already available off the shelf.

Can I now think in terms of a goal that I’m trying to achieve? I want a material to do a certain thing. Can I make it? That’s where we have the opportunity to change the way we think about making materials. Instead of using ready-made material, start with a goal and synthesize the material. That’s the excitement in the next 10 to 20 years. That’s what I’m getting at about the cyber physical world.

VB: Talking about the geopolitics of the tech industry, does that pull a lot of manufacturing back to the source, or to where it’s consumed? Do we still need all those giant factories in China, or do we just need 3D printers close to home?

Patel: Even if not at home, can we have small micro-grids of manufacturing? Even on the city scale of resources. This notion of having large-scale plants, whether it’s power plants, water, waste, or factories, this will give way to micro-grids. Not necessarily home manufacturing, but small-scale manufacturing of the type I knew when I came to the United States. I worked in a mulch factory in White City, Oregon. It was a small unit with about 20 people.

Economies of scale drove many large factories. But when you can right provision—in order to get economies of scale, I might print a million steel links, and I might not use 900,000 of them. They’re sitting on the shelves. But if we can give you what you want, when you want it—social networking and the sharing economy have shown that. They’ve created the abstraction layer, created the apps. Then we get to a world where if you want a steel link printed, a local shop in your neighborhood can make it for you. That’s what we’re going toward – micro-grids of small-scale manufacturing.

VB: Where do you think jobs are in the future, then? Running these micro-grids?

Patel: I’m always a believer that fundamentals never go away. Jobs are in fundamental skills – running the micro-grids, building the printing machines, whether ours or somebody else’s. When people ask me if we should get rid of the silos in engineering at the undergraduate level, I say absolutely not. Solid mechanics, mechanics of materials, that’s not changing. At the undergraduate level, keep the silos as they are. At the graduate level, create multidisciplinary curricula where you want to go.

The fundamentals we’ve had for generations don’t necessarily go away. They remain. As long as you’re strong in your fundamentals – physics, chemistry—trade skills are just as important.

VB: We have all the unicorns in Silicon Valley, the Ubers of the world. It seems like you’re also saying that the smokestack tech companies are very important.

Patel: Absolutely, in a cyber physical world. I was inducted into the hall of fame in 2014 and they asked me to give a talk. I said, “When I was a child growing up in India, I thought America was a land of tinkerers.” Children used to work on their parents’ cars. They delivered newspapers, made money, and bought radio-controlled planes. In India, when we wrote to our pen pals, no wonder those guys are ahead. They got to do things we didn’t have access to. My father had a Raleigh bicycle. There was no way we could work on a car. We were building gliders at best. They were steeped in fundamentals and they were hands-on. That’s why America was on top of the world. It was not a naïve perspective.

When I came to Silicon Valley I found myself in a valley of tinkerers. You’re sitting in the tinkerer’s home today. You saw the garage. Now I feel like we’re a valley of abstraction. We’ve abstracted the things below the internet. In the 21st century, the cyber age got us this abstraction. It was easy, because the abstraction is good. I like it, because you don’t have to know everything. But you can’t get carried away with abstraction and not question why, when you turn on this beautiful app, you also turn on a 100-kilowatt blower. You have to know how that blower works.

As long as you do the beautiful user interface and all that above the abstraction layer, and you know about the consequences at the bottom, then you’re good. But if you’re oblivious to the consequences below the internet layer, we’re in trouble.

VB: You can make your software 10 percent more energy-efficient, but then you might cause everything under that chain to consume a lot more energy.

Patel: Precisely. This cooling tower was exactly that. We wrote a paper and presented it in the data mining community. When you use a water-cooled tower, a water-cooled chiller in the data center, you save on electricity because you get more efficiency. Naturally people decided use that. But then you lose a lot of water through the cooling tower. Gallons and gallons vaporize. There’s always a balance. You have to know. Maybe you saved on electricity, but you wasted a lot of water. What does the compromise look like? What does the composite equation look like?

You’re absolutely right. That’s what happens. People optimize one thing, which is good, but forget everything else. Whenever I talk to software people—I learned a lot in software, continuous integration, continuous development. I’m trying to bring their best practices into hardware. I respect that. I expect the software team to also respect the hardware.

In the cyber physical world—I tell all the software people to make friends with mechanical engineers. That’s how we’ve created our teams. That’s exactly what we do with our affinity groups. When the machine learning community meets, we take on cyber physical problems.

VB: You’re working on the seams between cyber and physical?

Patel: Yes, doing co-design. I’m also talking to universities. I taught at Chabot College, a junior college in Hayward, for 16 years. I’m a product of a junior college. I see a lot of promise in our students. A lot of them were very well-placed, going on to do master’s degrees and so on. I taught at San Jose State, Berkeley, and Santa Clara. I taught graduate and undergraduate classes. I taught statics and dynamics and so on.

What I see in the last 20 years and the next 20 years, I feel like we have a lot of talent that needs to be directed. When I talk to universities, I’m thinking about—what’s the curriculum for the cyber physical age. That’s why I say the silos are important. The foundation must be strong. Then, at a master’s level, how do you think of a multidisciplinary curriculum?

It’s high time Silicon Valley looked at the supply side of human capital. We may be over-provisioned in one area, above the internet layer, but we’re under-provisioned below the internet layer.

VB: Didn’t we steer everybody that way because that’s where the money was?

Patel: Exactly. That neighbor of mine—I was having a conversation with my son. He said he wanted to go into civil engineering. My neighbor said, “There’s no hope. You won’t make any money.”

My son went into civil engineering anyway, at UCLA. He wrote in an essay, “The internet has created a coffin for American innovation.” He talked about when the eastern span of the Oakland bay bridge was coming up. None of his friends went to see it. He went to see it, because it was the opportunity of a lifetime to see a suspension bridge coming up. I said, “Well, maybe they saw it on Snapchat.” [laughs]

But there’s some truth to that. We’ve built San Francisco, with all these skyscrapers. There’s a lot of physical stuff around us. We hear stories about physical designs gone wrong. I sometimes wonder if we’ve become too cavalier around physical designs.

VB: I suppose the Internet of things is forcing people to start understanding physical objects again.

Patel: That’s a good thing. My daughter is a mechanical engineer. She was going to Tesla for an interview. I said, “Ask how they size the motor in the car.” Start with physical. You’re going to a cyber physical company. Start with physical.

I keep a whiteboard in the family room of our hours. The kids used to call it whiteboarding, a play on waterboarding torture. [laughs] But I said, “Personify yourself as the vehicle. Feel the wind. Calculate the wind resistance. Feel the rolling friction. Calculate the rolling friction. The motor must overcome rolling friction and wind resistance. At 80 miles an hour, what dominates? At 20 miles an hour, what dominates? At 80 miles an hour, how much power is drawn? Wind is dominating, but rolling is very low. With an 80 kilowatt-hour battery, how long will that last? You’re going to southern California in 115-degree weather and the air conditioning is on as you go up the Grapevine. How much does energy consumption go up?” All of these are fundamentals. There have to be actuators, too, because this guy Elon wants an autopilot. What about sensors? There’s a 77Ghz radar. We understood the physical platform. Then I said, “Okay, now go the cloud.” I didn’t go to the cloud until we actually understood the physical fundamentals.

You’re absolutely right. The internet of things exists, but we must make people understand the physical principles. You don’t have to design the motor. You can do an internet search and find out why they chose to use an induction motor. Children today can be much better-read than I was 30 years ago. The only thing is, we have to make our children go back and do the fundamental research.

I was inspired by the Beechcraft Bonanza, with its V-tail. I saw it when I was a child. I wondered why it had a V-tail, when no other airplane did. Now you can just go to Wikipedia on your phone and understand why. That’s what we have to instill, to go below the internet abstraction layer.

VB: Maybe space and Mars will inspire people again?

Patel: And once again, you think in terms of the movies. People were inspired by The Martian. It’s a powerful movie. A lot of these things, we actually can—as long as they keep asking the questions.

I’m very excited, given the opportunities. The counsel I always give to our community is depth in fundamentals, multidisciplinary perspective, and systemic innovation through multidisciplinary collaboration. I’m excited about being at HP for those reasons. There are many companies out there that I respect, but we have both cyber and physical.

VB: Do you have any different message when you’re talking to students versus talking to your engineers?

Patel: When I talk to students, it’s always the fundamentals. Well, even when I talk to engineers—I recently recorded a module for Brain Candy called “Energy for Everyone.” I talked about the first and second laws of thermodynamics. It’s the same. I don’t think there’s a difference. I always start with the fundamentals.