Since you joined GE Vernova last year, what are your top three priorities?
The three priorities are AI, robotics and automation, and they are connected. AI is delivering a return on investment (RoI) for us this year and it will grow exponentially. For me as CTO, it is also about launching new businesses that we call transformative programmes.
We talk about energy for AI and AI for energy. Energy for AI is a huge need as the base for AI is power. So what kind of best energy solutions can we provide — such as solid state transformers.
For AI for energy, it starts with operational efficiencies. For example, how do we make sure our quality processes are completely revamped with AI. What used to take eight to 10 weeks — sometimes three months — now takes two days.
We are also focused on how we can make our factories more efficient, improve product throughput and invest in robotics, where we will probably start to see significant RoI in 2028.
The third priority is transformative projects — businesses that do not exist in GE Vernova today. A few examples include a technique of catching carbon from the air and putting it in the ground that will reduce current emissions. This is known as direct air capture (DAC). It’s a research project right now, but we think it’s an opportunity for DAC to be a big product down the line. Another example is solid-state transformers for data centres which are far more efficient, flexible and capable than the current ones. The last one is fuel cells where they convert, which convert fuels into electrons through a chemical process.
There seems to be a capacity constraint for power across the world. Is it due to data centres’ demand for power?
The biggest bottleneck right now is energy for the data centres; the transformers, switchgear and power. For our gas turbines, we are taking orders for 2031. This is why we are expanding our capacity simultaneously. Our output is going to be four times more than what it was in 2019. At the same time, we are being judicious in capacity investment as the energy business is cyclical. The last time there was a big cycle in 2000, it went up and then it went down very quickly.
What you need for AI is data, chips and power. And data is not constrained at some level. Chips are a potential constraint because of geopolitical considerations. I think power is where the constraint is and we believe the infrastructure will happen. The only thing that could really change if chips become one hundred times more efficient and we do not need that much power. But we can see the roadmaps of all the chipmakers and while chips are getting much more efficient, we still need power.
What should be India’s energy strategy at a time when AI data centres are coming up?
It has become very clear in the last few months that energy security is a national security issue. It drives the economy and national defence. So having a secure energy source becomes really important. And that’s why countries like India are really leaning towards solar and wind because they’re not subject to supply constraints. So I think diversification of the source is going to be really big for India.
How does the research centre in Bengaluru fit into your AI plans?
Between the India centre and the one in New York, we conduct a lot of the core research for AI. We also have the AI Foundry, which develops AI solutions for us and customers. That team is located in the US and here. We don’t want it to grow to a very large team because it kind of builds the tools and sends them to businesses.
How is AI transforming your energy business?
There are three ways: Product development, quality inspection, and grid efficiency. Designing wind turbines and blades is very complex and one needs to look at thousands of designs. The process would usually take about 18 months, but with an AI technique called deep learning surrogates, it can be shortened to three months. The other part is field service. If we have to do a wind turbine inspection, we need to send two people for it. Now we use drones, which take pictures and advise us what to do. That helps us with turbine efficiency and bring down maintenance costs.
For quality inspection, we are using AI and robotics to inspect blades. The robot looks at it from all directions and figures out if there is a defect without relying on human judgement.
The last one is grid efficiency. Today, the grid is always designed for peak [usage] but the average is about 30-40 per cent less. There are a lot of software solutions that can make the grid much more efficient. It will also extend to how to keep the transmission lines cool when the temperature is hot.
