OpenSpace CEO Jeevan Kalanithi: 5 Questions

Not many proptech companies have been able to scale their technology to take advantage of the data center construction boom, but one that has cracked the code in servicing the sector’s gargantuan growth is OpenSpace.

The San Francisco-based visual intelligence platform has worked on more than 1,000 data center projects across the U.S., Europe, the Middle East, Canada, Australia and Singapore.  Overall, contractors and owners have used the platform to document and analyze progress on more than 100,000 construction projects globally.

Jeevan Kalanithi, CEO and co-founder of OpenSpace, spoke with Commercial Observer last week about what he sees as his company’s unique view into one of the biggest and least discussed challenges facing artificial intelligence: physically building the infrastructure required to support it. Kalanithi also discussed why the AI race has entered a physical phase, the biggest construction challenges emerging across hyperscale and AI-focused projects, why real-time visibility is becoming critical to managing risk and schedule certainty, and what owners, contractors and developers are learning as AI infrastructure scales globally.

This interview has been edited for length and clarity.

Commercial Observer: OpenSpace has worked with about 1,000 data centers globally. What does your company bring to servicing these unique facilities?

Jeevan Kalanithi: I think there’s two main narratives there. We just make it far easier and less risky to build out anything. Of course, we’re experiencing a boom in data centers, and our technology is super well tuned for taking the risk out of building those projects. 

The second thing we’re finding is that some of the latest advances in our technology and offerings are extremely well tuned for data, specifically what we call OpenSpace Track, which is our progress-tracking system. This is the ability to take information from all kinds of data sources — 360-degree cameras, mobile phones, etc. — which we’re really good at.

This visual way of working is just more efficient than the old way, and we can turn that into other insights an owner needs: Is this project actually meeting its schedule, is it actually happening? With the demand and price tags for these data centers, the cost of being late or making an error is insanely high. Recently I heard, and I won’t name names, that there was a data center delay and the liquidated damages for the delay was an incredible number — $500 million, which is absolutely mind-boggling. 

Obviously, like, you just can’t have that. So, if there’s new ways of working that make that less than 1 percent chance of happening, then that’s a huge difference.

You have said that AI has now entered the physical phase of construction. Is that being applied to data centers?

Let me define that. 

The current wave we are seeing with the foundation model companies is that they are transforming the economy. The reason why these data centers are being built out is to service that demand. The AI systems today that we all use are really trained on naturally digital information — text, spreadsheets, code and things like that. They’re incredible at that. But for the real-world economy that works in physical reality, the data that really matters is in the field. The  real data is visual, captured by cameras and things like that. When people use the term physical AI, they often mean AI systems that are capable of understanding the physical reality. 

Robotics is one, but another is managing the real-world economy, which in construction is a part of the systems that don’t just understand words but understand the physics going on. So, when we are deployed on data centers, our technology is really tuned to understand what’s going on there in reality, and it helps run better decisions more quickly.

Generally, proptech companies haven’t been big enough in scope to work on data centers in any manner. Is AI visualization the first foot in the door for the proptech industry?

I have my point of view based on OpenSpace and what we want to accomplish, and I certainly think that this visual-first way of working and spatial intelligence is absolutely being developed at the right time in the right place for these data center projects. It’s useful for just about anything you could build involving fact-based reasoning and collecting absolutely insanely massive amounts of data that these projects generate, and being able to make sense out of it. That’s what we do.

But, generally, there’s a lot of opportunity for technology to be used as part of this mega cycle of data center buildout. With data centers, it pulls into focus industry problems that have always been there, but really puts them into hyper focus, which is the mismatch between the amount of labor we have to build these buildings and the demand. You really need to find ways of making our workers more productive, because refilling the ranks is critical. I see that happening, but it’s not going to happen fast enough. 

To build out these data centers, given their price tags and complexity, you can’t go with the older ways of doing things and maybe be OK with the margin. So these folks are interested in trying to find new ways of getting these projects executed, because there’s no other option.

OpenSpace is labeled as a visual intelligence platform. How does it use its technology for data centers in particular?

It’s a pretty simple idea. The visual intelligence platform is pretty literal. What it means is building technology that is all about processing visual information. The idea of taking pictures of what’s going on has been part of builders’ existence since cameras were invented. 

Now, the big innovation is that with the technology we have today, we can use that at scale: take absolutely massive amounts of visual information, and point people at the right pyramid they need to make, as opposed to having petabytes of visual information or videos that people are scrolling through.

We’re applying AI for data centers, where one AI layer is kind of based on robotics that figures out where everything is at. The second thing is we have the human-AI hybrid system that turns it into high-level insights people typically make manually today for, say, deciding how to pay people. And the third is an agentic layer where you can design these agents to do various workflows that are time-consuming for you.

You believe that construction may ultimately influence the pace of AI deployment. Are you finding that with the explosion of AI facilities? Do you expect it to continue throughout the construction sector?

Yes. I would say that this visual way of working powered by physical AI makes sense to our customers. In fact, it’s powered by very advanced technology, but it’s almost like the old way of working where you see an issue, decide what’s going on, fix it and move on.

We’ve created a situation for builders where they’re drowning in digital paperwork, filling out a bazillion forms on their phone and all this stuff to log everything. It’s quite time-consuming. If we can replace that with just a visual look at something, log it and move on, it’s faster and easier.

We’ve been growing a lot across asset classes because of that natural resonance with the industry, and data centers are supercharging that in two ways. One, it makes these latent problems more acute, so the demand goes up. Two, these projects are absolutely massive. So, our deployments are larger. That’s been really good for us because we’re good at operating at scale.

The folks building out data centers are in such acute pain on getting these things done that they’re willing to invest and try new ways of working that I think were percolating through more slowly before. It’s really advancing the use of technology, and it had better work. 

They’re not going to do things that are pointless and waste their time. But I have found that they’re more willing to deploy technology more quickly because they’re under the gun. These projects can’t be late.

Philip Russo can be reached at prusso@commercialobserver.com.