From Goldman Sachs to Climate Investing in the AI Era

What we talked about.

In this episode of the Sunny Ray Show, host Sunny Ray talks with Rahul Parekh of 2150, a venture fund investing in technology for cities. Rahul traces his path from trading at Goldman Sachs, to investing and building at Rocket Internet, to founding his own business and now backing startups at seed and Series A. He explains the fund's thesis that cities concentrate economic activity, infrastructure and resource use, and that the physical economy has seen far less technological change than the digital one. Rahul describes how running a company changed the way he treats founders, why speed of execution must be balanced with safety and reliability in the physical world, and how climate benefits follow from better and cheaper technology. He uses portfolio company Gigaton, which applies AI to cement manufacturing, as an example of earning trust with industrial customers. The conversation also covers the shift toward distributed, intelligent energy systems, what 2150 wants to see from AI founders, and the risks even great founders cannot control.

Rahul Parekh explains why AI that acts in the physical world, not just answers questions, could reshape energy, industry and the cities we live in.

The questions, and the answers.

What is the thesis behind 2150 and what problem are you trying to solve?

Cities concentrate economic activity, infrastructure, resource consumption and innovation. They sit at the intersection of buildings, energy, transport and materials. These huge industries have seen little technological innovation compared to the digital economy, and improving them creates environmental and economic value. We now manage half a billion dollars, investing mainly at seed and Series A.

Which seat taught you the most about building a company: trader, investor or operator?

Being an operator, without question. Goldman Sachs gave me analytical discipline and a way to think about decisions under uncertainty, and Rocket taught me speed and execution. But being a founder and CEO was the most formative. It showed me how hard building a business is and gave me real empathy for the founders I back.

What did trading derivatives teach you about risk that you still use when looking at startups?

Risk is not just the probability of something going wrong, it is what happens if you are wrong. In venture the outcomes are more uncertain and the horizons longer, but the principle holds. We look for disproportionately large upside relative to capital at risk, and we separate fundamental business risks from ones a great team can overcome.

What did running a startup yourself change about how you treat founders?

I understand much better how hard the job is. As an investor it is easy to question a spreadsheet, but a founder manages hundreds of variables they do not fully control. I try to be the investor I would have wanted on my own board: available, direct, constructive, and someone founders can bring bad news to early.

How do the best founders in your portfolio find their first big customers?

They understand the customer's problem deeply. Industrial buyers will not purchase because technology is exciting, they need to believe in real economic benefit. Gigaton uses AI to optimize cement manufacturing, and it had to prove reliability and measurable savings first. Once you have that credibility, expanding and winning others becomes much easier.

Did you always see the built environment as where the big climate money would be made?

I would challenge the premise a little. It is not just buildings or climate. Much of the world's spending, energy use and employment sits in the physical economy, which runs on decades-old technology. The climate benefit comes from fundamentally better economics. If technology is better and cheaper, the sustainable option becomes the obvious choice.

What is 2150 looking for right now that it is not seeing enough of?

Transformational AI applications in the physical world. Many talented people are building around large language models, which is exciting, but we want AI that can control, optimize and automate real operations, like autonomous factories, intelligent energy infrastructure and supply chains. We want founders who understand both the technology and the economics of their industry.

Where does moving fast become reckless in the physical world?

Safety, reliability and trust are essential in physical systems, and you can never compromise on them. It is not only the model that matters but its integration into physical workflows. That is why these industries have been risk averse and slow to adopt software. AI now lets us optimize in real time, which we could not do before.