The real estate conversation around AI is focused on the wrong asset.
The real estate conversation around AI is focused on the wrong asset. Boardrooms are still debating where employees should sit. Meanwhile, the real constraint is sitting in a different building entirely. The data center is no longer just an IT procurement problem. It is the most consequential real estate decision in the modern corporate portfolio, and most executive teams do not own that reality yet.
During my 18 years at Deloitte, the last several working directly within the Office of the CEO, my eminence work focused heavily on the semiconductor ecosystem. What I understood then is the same thing I see every day now building Alderman Bennett: the semiconductor boom has a massive physical footprint. That footprint is the data center, and it is where technology strategy turns into a real estate, capital, and workforce constraint.
We have to talk about this because workforce planning means nothing if the physical infrastructure cannot support the tools your people need to use. Right now, most organizations are caught between two realities: scaling too fast for their infrastructure, or planning too slowly for the grid. Neither group is moving fast enough.
Before taking the West Coast Occupier Leader role at CBRE, and later launching Alderman Bennett, I gave serious consideration to a move into dedicated data center strategy. That led to a set of conversations in San Francisco with Tam Dell'Oro, the Founder and CEO of Dell'Oro Group. Her firm is not a peripheral voice in this market. Dell'Oro Group is the trusted source for strategic intelligence across telecommunications, networking, and data center infrastructure. The research her team produces is cited at the highest levels of corporate and investment decision making.
What I took from those conversations went well beyond market data. Liquid cooling sounds like a mechanical upgrade. Tam's team reframed it as a capital stack decision. It touches site selection, retrofit viability, construction sequencing, financing, and operating risk simultaneously. What I learned in those rooms I later watched play out in real time, working directly with occupier clients whose data center strategies were still catching up to the market realities her team had already mapped.
The multi-year AI expansion cycle is projected to drive worldwide data center capital expenditures to $1.7 trillion by 2030. Global spending is expected to approach $1 trillion in 2026 alone.
The Dell'Oro Group research makes the urgency impossible to ignore. The broader data center physical infrastructure market is forecast to reach $61 billion by 2029, growing at a 14 percent compound annual rate.
The primary physical constraint on all of that investment is heat. Most enterprise environments were built for single-digit to low-teens rack densities. Leading AI deployments are pushing far beyond that range, in some cases to 120 kilowatts or more. You cannot retrofit an older facility with better cooling fans and call it ready. The physics will not allow it. The liquid cooling market alone is projected to approach $7 billion by 2029. It is no longer optional. It is load-bearing infrastructure.
The deeper constraint — and the one most leadership teams have not priced in — is utility power. As of late 2024, nearly 2,300 gigawatts of generation and storage capacity were actively seeking U.S. grid interconnection. In many markets, wait times for new grid connections exceed five years. For leadership teams planning AI capacity, power is no longer a background assumption.
I have watched the CFO, the Head of Real Estate, and the CIO each work a different piece of this puzzle in the same organization, with none of them owning the whole thing. That gap is where companies lose ground they cannot recover.
This creates a timing problem that most leadership teams are not structured to solve.
Your strategy has to operate across three horizons, and they compress faster than most teams expect.
Securing power before the urgency arrives. Evaluate preleasing and phased commitments now. The goal is not to build a monolithic asset. It is to construct a hybrid portfolio blending on-premise, colocation, and hyperscale environments in a way that preserves optionality without surrendering capacity.
The talent and infrastructure collision becomes impossible to ignore. The industry has figured out how to hire engineers and coders. It has not figured out how to hire leaders who can govern intelligent systems at scale. But those leaders cannot govern systems that are physically bottlenecked by thermal limits and power constraints. The workforce problem and the infrastructure problem are the same problem. They just live in different budget lines. A company that hires the right leaders two years before the physical infrastructure can support them has not solved the problem. It has scheduled it.
There is no more runway for keeping the financial, real estate, and technology strategies on parallel tracks. Companies still running those functions separately will find their digital ambitions constrained by decisions that were made — or not made — years earlier.
There is no clean entry point into a market moving this fast. The discipline is not timing the perfect entry. It is committing early enough that the strategy still has options. The companies that lead this transition will not simply be the ones with the strongest models. They will be the ones that secured the real estate, the power, and the cooling to actually run them.
This is why I founded Alderman Bennett. Most organizations have the ambition. Very few have someone in the room who can connect the real estate, the capital structure, and the workforce implications at the same time.
Matthew Bennett Alderman is the Founder and Managing Principal of Alderman Bennett. Over an 18-year career spanning corporate strategy consulting at Deloitte and real estate execution as a principal investor, builder, and Senior Managing Director at CBRE, he built the practice to close the gap between enterprise strategy and the built environment. He founded Alderman Bennett to be the translation layer that does not exist inside any major firm. [email protected]
The Strategy Journal