AI is hitting commercial real estate hard, and if your portfolio isn’t being re-evaluated for it, you’re already behind. It’s not just a tech trend. It’s a fundamental change in how physical space supports business. JLL’s latest strategy says it all: they’re focusing on aligning portfolios to grab the upside of AI-driven growth while managing the new risks. So the real question for investors and developers is what you’re doing to get your assets ready for this.
Key Takeaways
- Use tools like Argus Enterprise for a data-driven audit of your assets to find the ones prime for AI infrastructure upgrades.
- Invest heavily in power and cooling, shooting for at least 200 watts per square foot for any data center or AI-focused office space.
- Get creative with leases, think flexible terms with clauses for tech upgrades and R&D space needs, not just old-school fixed agreements.
- Set up clear KPIs to measure AI readiness, think tenant AI adoption rates and energy efficiency, so you can actually track how your portfolio is doing.
1. Conduct a Granular Portfolio Assessment for AI Readiness
Look, you can’t just throw money at an “AI strategy” without knowing what you’ve actually got. I’ve seen too many groups waste a fortune because they skipped the basics. You need to get granular with every single asset, digging into its guts, the physical structure, the connectivity, the local talent. Then, you sort them into buckets: this one could be a data center, that one an AI research lab, maybe this one’s just a smart office. Without that map, you’re just guessing.
This isn’t gut feeling. Pull up your Yardi Voyager or RealPage data and get into the specs: floor load, power availability, HVAC. It’s simple physics, if a building’s floor can’t handle 100+ pounds per square foot, you aren’t putting heavy AI server racks in there. The real wins are finding properties that already have a good power grid and cooling, because trying to retrofit those later is where budgets go to die.
Pro Tip: Don’t plan for today’s needs. The demand for AI infrastructure is exploding, so what’s good enough now will be a problem in three years. Get electrical engineers in to look at grid stability and how much you can actually expand.
Common Mistake: Forgetting about fiber. AI needs fat, fast pipes for data. If your property doesn’t have multiple fiber routes from different providers, it’s a non-starter for any serious AI tenant.
2. Prioritize Infrastructure Upgrades for Power and Cooling
Everyone knows AI is power-hungry, but most property owners still don’t grasp the scale of it. We’re talking about data centers needing over 200 watts per square foot, which completely dwarfs the 50-75 watts in a typical office building. If you can’t supply that power, you’re not in the AI game. It’s that simple. A 2024 International Energy Agency report confirms the urgency, projecting that data center electricity use will double by 2026 from 2022 levels mostly because of AI.
So, the action item is to put money into upgrading substations, getting redundant power feeds, and installing serious cooling systems like liquid or adiabatic cooling that can handle these high-density racks. I tell all my clients: the bill for upgrading now is nothing compared to the cost of being left behind. While you’re at it, talk to utility providers about getting direct access to renewables, it’s a huge selling point for tech tenants who care about their sustainability story.
Screenshot Description: Here’s an AutoCAD schematic for a data center power upgrade. You can see the dual redundant feeds coming from the substation, the 2N UPS systems, and the liquid cooling for the server racks. This is what it looks like to take a rack from 100kW to 500kW.
3. Redesign Spaces for Collaboration and Flexibility
AI work isn’t siloed. You’ve got hardware engineers, software devs, and researchers all needing to work together, and the old-school office layout just gets in the way. If you want to attract these tenants, you need to redesign your spaces for how they actually work.
This means tearing out the cubicle farms and putting in flexible floor plans. Think movable walls, modular furniture, and shared “innovation hubs” with good AV gear. We’re seeing a real need for dedicated “war rooms” where teams can lock in for intense model training sessions, which means you need to provide large displays and a ton of network bandwidth. You’re building a space that encourages both random hallway conversations and intense, heads-down focus.
Pro Tip: Just ask AI companies what they need. Go talk to them. Many are hunting for very specific labs for things like robotics or sensor development, and these have unique demands that go way beyond a standard office fit-out.
4. Develop Agile Leasing Strategies
AI companies move too fast for long, rigid leases. If you’re still pushing 10-year terms, you’re going to lose them to someone who gets it. Your leasing strategy has to be as dynamic as they are. That means offering shorter terms, giving them options to expand or shrink, and writing clauses that let them bring in new tech as it develops.
You can even get creative with “flex-space” models or co-working setups designed for AI startups, maybe even throwing in access to shared compute resources or other expensive gear. I’ve seen some landlords experiment with revenue-sharing deals based on a tenant’s product success, but you’ll need good lawyers for that. The whole point is to lower the barrier to entry and reduce the commitment risk for these fast-moving companies.
Common Mistake: Insisting on a standard 10-year triple-net lease. It’s the fastest way to scare off a promising AI tenant. They know their needs will change dramatically in that time, and they’ll just go find a landlord who isn’t stuck in the past.
5. Invest in Smart Building Technologies and Data Analytics
An AI-ready building has to be a smart building. It’s about using your own building’s data to support the high-stakes work happening inside. By putting in IoT sensors, you can track everything from energy use to air quality, which generates a goldmine of data for running the building better.
When you feed that data into an AI-powered building management system (BMS) like Johnson Controls OpenBlue, you can start predicting problems before they happen, like an HVAC unit failing, which would be catastrophic for a room full of hot servers. This isn’t just about saving a few bucks on the power bill. It’s about providing the operational stability that an AI tenant absolutely requires. It’s a real feature you can sell them on.
Screenshot Description: This is a dashboard from a Siemens Desigo CC platform. It shows real-time power draw in a data hall, temps across server aisles, and even a forecast for potential cooling system strain. This is the kind of visibility we’re talking about.
6. Cultivate a Talent-Centric Ecosystem
At the end of the day, real estate is about where people want to be, and AI companies are fighting a war for talent. Your properties can be a secret weapon in that fight. This isn’t about adding a nicer gym. It’s about building an environment where top AI professionals actually want to work.
Location is everything. Are you near a university with a top AI program? Is there good public transit? Is the building in a neighborhood with a pulse? These are the questions your tenants are asking. A 2025 Gartner HR survey found that 70% of companies can’t find the AI talent they need, so a building that helps them attract and keep people is a massive competitive advantage. You could even go further by building out training rooms or partnering with schools to host AI workshops on-site.
7. Establish Key Performance Indicators (KPIs) for AI Readiness
If you can’t measure it, you can’t manage it. All these investments are meaningless without defined KPIs to see if your portfolio is actually getting “AI-ready.” You need to track specific numbers, like:
- Power Density Achieved: What’s the average watts per square foot you can deliver in your AI-focused spaces?
- Network Latency: How fast is the connection? Measure the round-trip time (RTT) to the big cloud providers.
- Tenant AI Adoption Rate: What percentage of your tech tenants are actually working on AI?
- Energy Efficiency: What’s your Power Usage Effectiveness (PUE) for the data center gear?
- Occupancy Rates of Flexible AI Labs: Are people actually using the flexible labs you built?
Look at these numbers constantly and be ready to change your plan. This market is new and changing fast. If an investment isn’t paying off, the data will show you, so listen to it and pivot. Don’t get emotionally attached to a strategy that isn’t working.
Getting a real estate portfolio ready for AI is a big, expensive job that demands a lot of forward thinking. But by auditing your infrastructure, rethinking your spaces, and getting smarter with your strategy, you can position your assets to be a core part of the growing AI economy and lock in their value for the long haul.
How much more power does an AI data center really need?
A lot more. AI data centers need huge power densities, often over 200 watts per square foot, because the computing is so intense. Your standard data center might only need 50-75 watts/sq ft. That’s why the infrastructure upgrades are such a big deal.
Why is everyone talking about cooling for AI buildings?
Because it’s absolutely essential. Those powerful AI servers throw off a massive amount of heat. Without advanced cooling systems, like liquid or adiabatic cooling, they’ll overheat, fail, and throttle performance. Bad cooling means you can’t run the equipment properly.
What kind of leases do AI companies want?
They want flexibility. Think shorter lease terms, options to easily expand or shrink their footprint, and clauses that let them upgrade their tech. “Flex-space” or co-working deals where they can get access to shared, expensive computing gear are also a big draw.
Does location still matter for an AI portfolio?
It matters more than ever. You need to be near top universities with AI programs to tap into the talent pipeline. You also need rock-solid infrastructure: redundant fiber optic networks and a power grid that can scale. Good public transit and a decent neighborhood help your tenants keep the talent they hire.
How do smart building features help with AI tenants?
They provide operational stability. Using IoT sensors and an AI-driven building management system gives you the real-time data to optimize energy, predict when equipment might fail, and keep the environment stable. For a tenant running millions of dollars of hardware, that stability is a key selling point.
