Ep.79 Rigour, creativity and systems thinking: a Silicon Valley veteran’s insights

Maria Rampa: Hi, I’m Maria Rampa, and welcome to this episode of Engineering Reimagined, recorded live at the 2025 CAETS Conference in partnership with the Australian Academy of Technological Sciences and Engineering, or ATSE.

In today’s episode, Aurecon’s Chief Engineering, Eminence and Innovation Officer Tanya De Hoog interviews Silicon Valley pioneer and Hall of Fame engineer Chandrakant Patel.

Chandrakant is a pioneering Silicon Valley engineer with over four decades at HP Inc., where he served as Chief Engineer, Senior Fellow, and head of HP Labs. A leader in energy-efficient computing, AI, and sustainability, he holds 167 patents and founded HP’s Smart Data Center research program, which helped shape multi-billion-dollar infrastructure solutions.

Before going on to become a Silicon Valley engineering veteran, Chandrakant grew up in India at a time when there were no screens, no digital design tools, and in fact, not even television. It was a time when imagination was everything. A passionate cricketer, he painted vivid pictures of the Gabba or MCG in his mind after intently listening to the crackling voice of a commentator. He never saw Dennis Lillee bowl, but through words alone, he could picture the wind, the dew, the anticipation.

This power of imagination didn’t just bring the game to life for Chandrakant, but also laid the foundations for a lifelong career in engineering. At its core, engineering is about imagining what doesn’t yet exist, and finding a way to make it real.

In this episode, we explore how imagination fuels innovation and why timeless engineering fundamentals like rigour, creativity, and systems thinking are more important than ever.

Chandrakant also explains why he believes the future belongs to those who stay curious and keep learning, and how data centres of the future might run on renewable and even agricultural waste energy.

We hope you enjoy the conversation!

+++++

Tanya de Hoog: Hello and welcome to Engineering Reimagined, recorded live at the international Council of Academies of Engineering and Technological Sciences, or CAETS Conference here in Brisbane. I'm Tanya de Hoog, and today, I'm joined by Chandrakant Patel, inducted into the Silicon Valley Engineering Hall of Fame for his many contributions. He's a recognised leader in AI, energy efficient computing and sustainability. Welcome Chandrakant.

Chandrakant Patel: Well, thank you for the kind introduction, Tanya.

Tanya de Hoog: How about we start with our shared passion for engineering? So, I just learned that both your father and your son are structural engineers? So we already have some things in common, and just looking at all of your achievements, one of the things that stands out to me is creativity and innovation in engineering. One of the things that I hold dear to my heart in engineering terms is core engineering attributes. Those things such as rigour, systems thinking, subject matter expertise and of course creativity and innovation. So, coming from Silicon Valley, which is synonymous with innovation, love to hear a little bit more about your perspective on where you started as an engineer and what engineering was like then and what your perspective of the future of engineering might look like.

Chandrakant Patel: I want to start by saying that the attributes you described, fundamental, systematic execution, are absolutely the underpinnings of where we are going.

Tanya de Hoog: That's great to hear.

Chandrakant Patel: I'm convinced of that. I don't believe that I would have had the success had I not had that rigour. And that rigour comes from the structural engineering influence that I've had from my dad. You will notice that even my most complex projects, like a large-scale data centre, start with a simple sketch where I describe the entire system from the source of power at a hydroelectric power plant, losses of available energy going into the server. Heat going out to the cooling tower. I draw the big picture first and that comes from the structural engineering thinking. So big picture thinking, holistic thinking, systematic execution, underpin the age we are entering. So let's start from your question. So 42 years ago, I started in Silicon Valley. My first day on the job I was on a drafting table and I designed hard disc drives, the early disc drives. The floppy drives were 1.4 megabytes. We were building a hard disc drive, it was the size of a washing machine, nine discs, 32 heads, two voice coil motors, a gigabyte of storage, sold for $50,000. The head disc assembly was 40 kilogrammes. There's two things that happened in Silicon Valley at that time. I did all the design using drafting table. CAD had not arrived yet, there were mainframes. That's the first. Second, was everything was vertically integrated. You source the discs locally. You coated it with magnetic material. Manufacturing was local. It was fully vertically integrated. That's the Silicon Valley I started. The same is true for chips. Chips were designed. The layout was done on floors, on acetate paper, and the chips were locally manufactured. That was the Silicon Valley of the 80s. And over the years, we went into computer-aided design. I taught at a community college, both engineering graphics and CAD. And whenever I taught, I would always still keep a drafting table on the right-hand side for conceptual drawing and a CAD workstation on the orthogonal to the drafting table, so the grammar of drawing is not lost. Then we entered CAE, computer-aided engineering. And on the electronic side, we went into electronic design automation. So now a lot of the designs were automated. You would give the points that you want to wire the chips to, and it will automatically wire, and so on and so forth. So a lot automation came about. So that was the big change that occurred, how design was done. And then the Valley went from being vertically integrated to more like sourcing components from all over the world and systemically integrating it. And the Valley entered what I would call the cyber age. That was the age above the internet abstraction layer. And the Valley, I think, focused more on human-generated data, e-commerce, social media boom. It was mostly how do you connect people and how do you use people information to build solution largely above the internet abstraction layer. Now we are entering the cyber-physical age, where we cannot just sit above the abstraction layer called internet. We need to go below the abstraction layer and understand the hardware, the physical side, whether it is power, water, waste, transport. And now the conversation must shift from IT stack, which is stack of cyber technology, to cyber-physical stack, which is the stack starting with domain knowledge, fluid mechanics, heat transfer, actuators, sensors, all the way up to software, hardware, software co-design. So we have entered the cyber-physical age. The fact that we have entered the cyber-physical age suggests that we cannot forget our fundamentals, because cyber-physical is built on the domain fundamentals on the physical side. These are complex systems, so you have to be systematic. Let me make a connection to civil structural. When you do seismic analysis, when you dot the i's and cross the t's, so the buildings would last for hundreds of years, that level of rigour is going to be needed in the cyber-physical age and is needed in this cyber-physic. So the way I tell the youth is, stick to your fundamentals, know your faculty, know the textbooks you use, and of course, you use all the AI tools and all the tools that are necessary to be expeditious, but don't leave home without your fundamentals.

Tanya de Hoog: Tell me a little bit more about innovation and creativity in terms of whether you think they're fundamentals of engineering or something that you develop over time and how important are they to the future of engineering?

Chandrakant Patel: So there's two things to it, I always say. One is the fundamentals of engineering. But before I went into engineering, I was a child growing up in India. And back in the day, there was no television. So, I would imagine Gabba or Melbourne Cricket Ground through the eyes of the commentator. We would be sticking to every word. He'll be talking about the wind blowing, the dew on the ground, and how it would affect the bowling. We would imagine Dennis Lillee walking up to his bowling mark, and here's the world's fastest bowler. Never saw Dennis Lillee, but we imagined. That imagination would go on to help me in my career. So, I think engineering is about imagining things. It is easy in today's day and age to forget that. So how do you create an environment where people would imagine? We were forced to imagine because we didn't have all these tools available.

Tanya de Hoog: One of the common questions that I'm asked as Chief Engineer by early career professionals is, 'Is my job going to be replaced by AI in the future? And do I have a future as an engineer?' My personal view is that if you want to make a difference in the world, engineering is the place to do it, in this moment and will be. We're stepping into this age of the engineer. How do you approach that question knowing a lot more than I do about AI, knowing a lot more about some of those fundamentals that may be replaced?

Chandrakant Patel: So I distinguish AI between AI, which is built on the digital footprint of all the people. We gave all our information on the web and it's been scraped, and people sell us goods and people connect us. Fine, that is social media, e-commerce. But when I think of AI, I think of physical AI, AI that enables solutions that matter to society. Solutions that are addressing the challenge of perturbation between supply and demand. So on the supply side is a pool of resources. On the demand side are our basic needs, power, water, waste, transport. There are these trends that are causing a huge problem between supply and demand, and those need to be solved. And if I even take one trend, I see the opportunity is so huge in physical AI. Physical AI has hardly scratched the surface. So the opportunities are so huge that people with domain understanding will be the king. So why? Because physical AI necessitates domain knowledge. And I'll give you an example. The collapse of Tacoma Narrows Bridge, vibration, for example. So if you have a train which is going at high speed, oscillations occur. The ability to find, detect an anomaly from machine-generated data would exist, but finding the root cause of anomaly requires domain knowledge. So in order to create solutions using broad AI, where you're collecting machine-generated data with context, analysing it, inferring and acting. You need people with domain knowledge. So physical AI with machine-generated data requires domain knowledge so we can build the inference engine. It'll be expeditious, but the building of it will require domain knowledge. The second part of AI is what I call domain-specific generative AI, where you have engineering workflows, where you have a chatbot, you ask questions. But those prompts need to be steeped in domain knowledge. You could teach prompt engineering and tell people how to structure the prompt. But the prompts themselves have to have an engineering framework. So all those who feel that AI will replace their job, particularly in engineering, in hard engineering, let's call structural engineering, hard engineering, chemical engineering, mechanical engineering, complex engineering of automobiles, power, water, waste, transport, healthcare, I don't think their job goes away. You enable AI tools that will solve problems that matter to society. AI is going to have to pivot from frivolous AI to AI that matters to society.

Tanya de Hoog: So as engineers, we're okay, but maybe what I'm hearing from you is the value of those fundamentals and that rigour and that systems thinking coupled with creativity and innovation and deep domain knowledge are the things to lean into for the future of engineering.

Chandrakant Patel: I am 65 plus, I'm learning every day. So, I also encourage the same people who are focused on fundamentals in physical sciences, to also learn cyber sciences. So, I go and learn online, what is the latest in AI, what tools can I use? I experiment with those tools to create engineering workflows that are domain-specific generative AI. That way I experiment with it, get excited about it, don't be afraid of it.

Tanya de Hoog: Lifelong learning rather than the fear of it. One of the things that I get excited about is, what's the untapped potential in engineering with new tools, with access to AI in different ways in the future? What are those problems that we haven't been able to solve in society or from an environmental perspective, that might now be possible when we couple engineering from the past with tools of the future?

Chandrakant Patel: Suraj, my younger son, coincidentally, after his wedding, I was taking his guests on a boat ride underneath the Golden Gate Bridge. When I went underneath the Golden Gate Bridge, I asked, if I were to build the Golden Gate Bridge today and I used all the tools, I used domain-specific generative AI to expedite my engineering design, I would use CAD, CAE, and my fundamentals to do the design work. Will I be able to build the bridge in less time? And more importantly, I use joules as the currency. I don't use dollars, rupees, or yuan. So unit of energy, joules, and available energy from second law of thermodynamics. So from that point of view, if I were to use AI, would I be able to build the Golden Gate Bridge in the same time, but with less joules, lifetime joules? So cradle to cradle, extraction, manufacturing, operation, use, and reclamation. So would I use cradle to cradle, less joules with AI? And that's what I'm excited about. If AI uses more joules, then what's the point? AI must deconstruct conventional business model and replace them with AI-based model that take less energy. So what is often missing in AI is, what is the metric? So my metric is very simple, what I call net positive impact. So if I use AI, I should be able to build a system like the Golden Gate Bridge with less energy that it took with the conventional model, inclusive of AI energy that is needed to run the computation. AI must deliver net positive impact. Unfortunately, when I think hard, I'm not sure that we can build the Golden Gate Bridge in the same time with less joules. I'm not convinced.

Tanya de Hoog: So we've talked a lot about engineering fundamentals and AI, but of course one of the things that you're a specialist in and have been from early days is data centres. With the increasing demand for data centres, because of the evolution and the rapid pace of AI, what do you think the future is for data centres?

Chandrakant Patel: So I think the future, first we have to start with the data centre design. How should we think about data centre design in light of the fact that data centres consume a lot of available energy? Water, I also consider available energy. I convert water to joules. So I take water, and I say then I will assume that nature desalinates, but it takes joules to move the water, treat the water. So I think of the entire cost of data centre as joules. So as data centres' lifetime joules are so high, particularly with the rise of AI, one must start to ask yourself, how should we think about data centre design? And I think of data centre from a supply side and demand side perspective. From a supply side perspective, I have three principles. One, least lifetime energy from cradle to cradle for the data centre. What are the ways by which I will design the data centre, so I minimise the energy consumed from a cradle to cradle perspective? Principle number two, local sources power the data centre by local sources of available energy. So what are those? Sun, solar could be one. Another is look at waste as an opportunity. Because our third principle is, look for available energy in waste streams. So, let's say one joule at 500 degrees celsius coming out of a chimney in a factory still has 0.6 joules available. Can we use that waste stream? Another waste stream is dairy farms, manure from dairy cows. A dairy cow produces 55 kilogrammes of manure. 2,000 dairy cows would support a one megawatt data centre. So the manure gets anaerobic digestion to create methane. Methane is used to drive a generator and the electricity is used to power the data centre. Waste heat from the compute racks would be sent into the anaerobic digester because the manure has to be kept warm to enhance methane formation.

Tanya de Hoog: So you're thinking systems already and you're thinking about fundamentals.

Chandrakant Patel: Exactly, so what I'm saying on the supply side, local sources of available energy, including waste streams, and building that power plant, the anaerobic digestion of manure to generate methane is a firm source. Solar is a diurnal source. So the output will be a solid firm source plus diurnal output. So that's the pool of available energy I'm going to give Tanya. Now Tanya builds a data centre. The demand side of the data centre should be managed so that you stay under that curve. I'm not going to give you an infinite amount of electricity. You will schedule your jobs. If it is not important, you look at the service level agreement and schedule the jobs when the sun comes out. Other jobs that are important, you run it on methane from anaerobic digestion. You may use natural gas powered solutions. So you have to have a local power plant with demand and you manage the demand based on the supply. That will force you to think systemically. Then once you have gone through the systemic thinking, you go inside the data centre and go about building control systems. And it's a multi-input, multi-output problem. There are thousands of servers. There's fluid flow, whether it's liquid cooling or air cooling, has lots of actuators. So it becomes a multi-input, multi-output problem, it becomes a complex problem. You start with domain knowledge and use AI to build a plant function. So you could do a closed loop control system and you schedule things, the cooling, the power based on the need, need-based provisioning inside the data centre using sensing and control, which will require AI because it's a very complex system, but not without domain knowledge. Starts with domain knowledge, then AI. So what I'm suggesting is integrated supply-demand management. So with that kind of blueprint in mind, you got to ask yourself, what is the data centre of the future? And you see where AI is going. So AI today has been built on data that's available, at least generative AI has been built on data that is available on the web. But there is a huge opportunity in domain-specific generative AI. So power, water, waste, healthcare, your own field of structural engineering. Surely you didn't put all your private data out on the net. So if somebody wants to build the tools, they need to come to Tanya and say, I need to buy your data. Shouldn't Tanya monetise your own data? So when I take the train here, when I run the ferry, I'm perplexed by how quickly the crew docks the ferry, gets out, I ask myself, there's so much data, so much information, so much institutional knowledge. That institutional knowledge does not belong on the web. That must be harvested. And that's just in one sector. Think of healthcare. Think of all the private and medical data here. All that is proprietary data, there is a lot of domain knowledge. In order to service that, to create an AI built on proprietary data and proprietary domain knowledge, we will see the rise of local data centres in Australia, a network of data centres. And when you are dealing with machine generated data, the data centre need not be campus, large-scale, warehouse-scale. They may be more modular data centres and for security, they may be underground. You don't want these data centres to be sticking out like a sore thumb from a cyber-physical security. So I think the next chapter in data centre is going to be built on domain knowledge, applying first principles, second law of thermodynamics. It'll be focused on proprietary data, and it's an opportunity for those with proprietary data to monetise it. It's an opportunity that won't come again. And that's why I think we are soon to see yet another phase of data centre beyond what you're already seeing with AI data centre. And that’s what I'm excited about.

Tanya de Hoog: When we work on something as an individual, we can have quite a profound effect. But when we share that knowledge, and we share that beyond just the people that are specifically interested, into systems that can be adopted, actually that impact can be quite profound. So, wrapping up, with that in mind, if you were starting your career in this moment as an engineer. And you were purpose-driven from a net positive perspective, what would you do? Where would you invest your time and your efforts to have the greatest net positive impact?

Chandrakant Patel: I got goosebumps because when you said the sharing. In my career, the best advice I got was from my first manager at HP. He said, “Chandrakant, share and move on.” What he meant was, I was doing dynamics of structures and disc drives, and he said, “Learn that. Learn by doing, teach others, move to the next.” I went from disc drive design to chips. I had never seen silicon before but I would go on to do flip chip, multi-chip modules, CPU complexes, and so on, right? To proprietary computer systems, to data centre. I kept pivoting, pivoting, pivoting. The way I pivoted was whatever I learned, I taught somebody else. We have a lot of youth coming to the CAETS this time through the Elevate Programme. My advice to the Elevate Programme is share and move on. Don't keep it to your chest. Move to the next. The one thing I learned in my career, and I would go on later to create a career strategy seminar that I run, it's called the Visual CV, Visual Curriculum Vitae, where I actually articulate my CV by sketch. On one axis I have my depth, another axis I have my industry subject, and on another axis I have my breadth subject. So when I went from disc drive to chips, what breadth subjects do I need to pick up? And that way I became a T-shaped contributor. T as in Tanya. If your depth is in history, keep that depth. Nurture that depth but pick up breadth. So think systems, become T-shaped, with depth in one, breadth in many, and never stop learning. So that's what I would do, but not T-shaped at the expense of losing the fundamentals in your depth subject, because often times people think of T-shape as, oh, I should have breadth in many. No. You must continue to garner depth. In fact, I'm paranoid about forgetting my depth in mechanical engineering to the point I keep teaching courses from undergraduate, kinetics, kinematics, dynamics of structures. I'm afraid that I would lose the knowledge I gained in undergraduate in Berkeley, 1980. You know, Den Hartog's book in mechanical vibrations, Egor Popov in mechanics of solids. I keep that close, and I always told my kids, no PDF in my house, you'll have textbook in my house, because the learning you'll garner from textbook, you will never forget.

Tanya de Hoog: Well, look, this has been wonderful, and we could talk for hours. We have so much in common in terms of our shared values, but it's incredibly inspiring to talk to you. Thank you for sharing.

Chandrakant Patel: Thank you for having me. It's an absolute honour.

+++++

Maria Rampa: I hope you enjoyed this episode of Engineering Reimagined, with insights from industry veteran Chandrakant Patel about how engineering fundamentals like rigour, creativity and systems thinking are as important today as they have ever been.

If this episode has inspired you to keep listening to our podcast, hit subscribe on Apple or Spotify and don’t forget to follow Aurecon on your favourite social media platform to stay up to date and join the conversation.

Until next time, thanks for listening.