By Hamza Faisal
Tenants renting space in artificial intelligence data centers have sharply raised the price of being late. According to a report from The Information, cited by multiple outlets this weekend, tenants negotiating new leases are now demanding more than $5 million for every day a developer fails to deliver power and finished facilities on time, up from about $1 million a day as recently as this spring.
The fivefold jump in a matter of months captures the new reality of the AI building boom: computing capacity is the prize, and the companies that need it most are no longer willing to wait quietly for it.
From $1 million to $5 million in months
The shift is visible inside the insurance market. Mike Mathews, who runs the global digital infrastructure practice at insurance broker Marsh, had his team put together a custom delay-in-delivery policy this spring for a developer whose tenant wanted $1 million a day if the first phase of a gigawatt-scale campus missed its date. Within months, tenants in fresh deals had moved the demand to $5 million a day and beyond.
Marsh has been building for this moment. Its Nimbus facility, launched in June 2025 for large data center construction, started with up to $1 billion in construction all-risks limits and $365 million for delay in start-up. In January, Marsh raised the facility's limits to as much as $2.7 billion, a sign of how quickly the sums at stake have grown.
Why tenants hold the leverage
The tenants driving the tougher terms are mostly hyperscalers and AI laboratories, the deep-pocketed buyers every developer wants locked into a long-term lease. Their leverage is showing up directly in contracts, with penalties joined by rent abatements, liquidated damages and, in some cases, termination rights tied to when power and buildings are actually ready.
The urgency is grounded in a difficult delivery outlook. Goldman Sachs expects only about half of scheduled United States data center capacity to arrive on time, while other coverage puts the on-time share at 50 to 60 percent over the next one to two years. Power availability, permitting, equipment shortages and construction delays are all cited as constraints, with power the main obstacle.
One deal shows how sharp the terms have become. Under SoftBank-backed SB Energy's arrangement with OpenAI for a 10-gigawatt campus in Ohio, OpenAI receives one free day of base rent for every day of delay, and that penalty doubles once the delay passes 90 days. A full year late would cost 640 days of free rent, close to two years of income on the project.
The risk now sits with builders
Developers are already feeling the pressure in other ways. Oracle issued a force majeure notice in September 2026 on its Project Jupiter campus in New Mexico, preserving its rights to defer rent if power delays affect the planned 2028 opening, even as the company says the project remains on schedule.
The message for the industry is blunt: in the race to build AI infrastructure, the schedule is now part of the price. Tenants are transferring more of the risk of late delivery onto developers, and insurers are rushing to cover a risk that, until this year, barely had a market at all. As long as power and equipment remain scarce, the cost of opening late is only likely to climb.
