This is a straight look at a hard truth that too many AI conversations still try to dodge: the boom is no longer abstract. Referenced from the reporting by Joshua Mann, the current wave of data center development is capital deployment at Industrial Revolution scale, only this time the factories are measured in megawatts, coolant loops, substations, and labor scarcity. The story is not just that demand is rising. It is that every layer of the physical system is being pushed at once, and the strain compounds. Costs are climbing, project requirements are changing, and the people needed to build this infrastructure are becoming the least hedgeable input in the entire stack.
The Scale of the New Buildout
There is a tendency to talk about AI as if it floats above the ground. It does not. It sits on land, drinks power, throws off heat, and requires an astonishing amount of physical coordination to exist at all. That is what makes the current moment feel so much bigger than a standard tech cycle. This is not another software land grab. It is a construction and utilities story wearing a technology headline.
Joshua Mann’s framing gets at the core of it: this buildout resembles an industrial revolution because of the sheer concentration of capital now rushing into real assets. Not nice-to-have assets. Mission-critical, power-hungry, time-sensitive infrastructure. And once that kind of capital starts moving, it pulls on everything around it. Grid access. Equipment manufacturing. engineering capacity. Skilled trades. Rural logistics. Permitting timelines. Water strategy. The whole machine.
That is why the market feels so unforgiving right now. Demand is not rising in one neat lane. It is hitting every lane at the same time.
The Cost Curve Is Already Telling the Story
The data is blunt. According to the Turner & Townsend Data Centre Construction Cost Index, average data center construction costs rose 9% in 2024 and another 5.5% in 2025. Those are not background fluctuations. That is a sector under pressure.
And that pressure matters because these projects are already expensive before you add the AI-specific requirements. A few points of annual escalation at this scale is not a budgeting nuisance. It is a strategic problem. It changes underwriting. It changes schedules. It changes where projects pencil out and where they die.
A lot of executives still treat construction inflation like a temporary headache. That feels naive. What these numbers suggest instead is a sector resetting to a new baseline, where demand for capacity is so intense that costs do not just rise. They stay elevated because the supply side cannot expand cleanly enough to catch up.
The AI Tax Is Real
Here is where the conversation gets sharper. AI workloads are not just increasing demand for more data centers. They are changing what a data center has to be.
That creates what many operators are already living through as an AI tax. Liquid cooling, once treated like specialty infrastructure, is becoming standard for high-density AI deployments. And it is not free. The premium for liquid cooling systems is running about 7% to 10% above traditional approaches. That matters on paper. It matters even more in the field, where the premium is not isolated to one line item.
Because the real issue is multiplicative strain.
Add liquid cooling and you are not merely swapping one thermal system for another. You are changing design assumptions, procurement sequences, commissioning complexity, maintenance requirements, and staffing needs. Push rack density up and you stress power delivery. Stress power delivery and you start running into grid realities. Run into grid realities and suddenly onsite substations move from optional to necessary. Then those substations have to be sourced, transported, integrated, and staffed around. Every fix creates the need for another fix.
That is the multiplicative part. AI infrastructure does not scale in a straight line. It scales by forcing adjacent systems to upgrade with it.
Why Human Capital Is the Hardest Constraint
Materials can be hedged, at least to a point. Developers can forward-buy copper. They can lock transformer orders early. They can negotiate equipment slots. They can stockpile certain components if they have the balance sheet and storage plan to do it.
People are different. You cannot warehouse a labor force.
That is the part too many spreadsheets still flatten. The skilled workers needed to build, install, test, and maintain this new generation of infrastructure are not interchangeable, and they are not instantly scalable. High-voltage specialists, cooling system experts, commissioning teams, controls technicians, and experienced project managers are all under pressure at once. When demand surges across multiple markets, you do not just pay more. You wait longer, compete harder, and accept more execution risk.
And that is why the human capital factor deserves its own category, not a footnote under labor. A project can survive price volatility better than it can survive missing expertise. Miss on materials and you may absorb cost. Miss on people and the whole schedule can buckle.
This is where Joshua Mann’s point lands hardest. You can hedge commodities. You cannot hedge whether the right technician is available in the right place at the right time to get a mission-critical system online.
The Physical Requirements Have Changed
The old mental model of data center development is already out of date. Facilities designed for conventional enterprise loads are not a useful benchmark for what AI infrastructure now requires.
One of the clearest shifts is the growing need for onsite substations. For large-scale AI loads, reliance on legacy utility connections is often not enough. The power demand is too steep, too concentrated, and too immediate. So the project expands. It becomes part building, part energy infrastructure, part logistics campaign. That means more land considerations, more utility coordination, more specialized design, and usually more time.
Then there is the geographic wrinkle. Some of the most viable sites for hyperscale development sit in remote areas where land and power access are more workable. Great on paper. Messier in practice. Remote labor markets are thinner. Housing for workers can be limited. Travel times are longer. Specialty subcontractors may need to be brought in from far away. Maintenance staffing after construction can become its own long-term challenge.
This is one of those details that sounds operational until it starts wrecking timelines. Remote-site development is not just a real estate decision. It is a labor strategy, a logistics strategy, and a retention strategy all at once.
The Bigger Point
The cinematic version of AI is all inference and intelligence and software magic. The real version is concrete, steel, coolant, switchgear, and people who know how to make all of that work under pressure.
That is why this moment deserves to be treated with the seriousness of an industrial shift. The headline numbers, the 9% cost increase in 2024 and 5.5% in 2025, are not random spikes. They are markers of a market being physically re-engineered in real time. The liquid cooling premium of 7% to 10% is not a niche surcharge. It is part of a broader AI tax that ripples across the entire development chain. And the labor challenge is not secondary. It may be the main event.
The companies that understand this early will act differently. They will stop treating AI infrastructure like a faster version of legacy data center construction. They will plan for substations earlier, source talent more aggressively, evaluate remote-site realities with less wishful thinking, and price the multiplicative strain honestly.
That is the real frontier here. Not just building more capacity. Building it in a world where every layer of the physical stack is tightening at once.
Built to grow strong businesses, meaningful partnerships, and lasting community impact. Connect with McFadden Finch Holdings Company today.
McFadden Finch Holdings Company
Vision. Leadership. Lasting Impact.
Lake Merritt Plaza
1999 Harrison Street, Suite 1872-73
Oakland, CA 94612
(510) 973-2677
www.m-fhc.com
info@m-fhc.com
McFadden Finch Holdings Company (MFHC) is a premier holdings and investment management firm dedicated to driving sustainable growth and long-term value. Our mission is to bridge the gap between visionary capital and community-centric development, ensuring tomorrow’s infrastructure meets today’s needs. Through strategic project management and rigorous market analysis, we empower our partners to navigate the complexities of the California economic landscape with confidence and clarity.
For more information on how MFHC can support your industrial or real estate investment strategy, contact us at (510) 973-2677 or visit www.m-fhc.com.
Sources
Based on the story by Joshua Mann in the Business Journals.
Referenced from the Turner & Townsend Data Centre Construction Cost Index 2025–2026.
Additional data sourced from S&P Global.
Disclaimer: This content is for general informational purposes only and does not constitute legal, financial, tax, investment, real estate, business, or other professional advice. Reading this content does not create an advisory, client, fiduciary, or contractual relationship with McFadden Finch Holdings Company. Because every business, investment, property, and strategic situation is different, you should consult qualified professionals regarding your specific circumstances. McFadden Finch Holdings Company makes no warranties regarding the accuracy or completeness of this information and is not responsible for third-party content, links, products, services, or organizations referenced. Testimonials, examples, case studies, and projected outcomes are illustrative only and do not guarantee similar results.
