The race to expand artificial intelligence infrastructure is fueling one of the most aggressive data center development booms in decades—but investors and analysts can’t see exactly where the money is going. According to The Wall Street Journal, as AI workloads are projected to rise by at least 20% in 2026, major tech companies are pouring billions into construction projects that remain largely opaque on their balance sheets.
These companies are seeking new data center sites and massive quantities of specialized chips to power AI processing, but few details are disclosed about how much each piece of the puzzle costs. As the Journal explains, companies treat data centers and their high-powered equipment as long-term construction projects without breaking out individual expenses, leaving investors to guess at the financial risks behind these enormous bets.
The challenge extends beyond accounting. The physical lifespans of AI infrastructure vary sharply—buildings may last for decades, but servers and chips can become obsolete within a few years. When next-generation hardware demands new power and cooling systems, that turnover can even shorten the useful life of the buildings themselves.
“The construction-in-progress account is this big hole where hyperscalers can bury a lot of their costs,” Gaurav Kumar, an accounting professor at the University of Arkansas at Little Rock, told the Journal.
Data center development costs are now surpassing those for office construction, driven by intense demand for digital capacity. Yet new grid connections and power infrastructure can take years to complete, raising the risk of delays. As the Financial Times has reported, developers are scrambling to secure future electricity supply—often for projects that might never be built—making it harder for utilities to separate real demand from hype.
For investors evaluating these projects, the lack of financial transparency poses significant challenges. Predicting cash flow is difficult when tech companies can suddenly shift plans in response to fast-paced AI innovations. “The disclosure is not evolving quickly enough to keep up with real-life demand for information about AI investment,” accounting consultant and journalist Olga Usvyatsky told the Journal.
Though some companies have suggested they’ll extend the lifecycle of server equipment to boost efficiency—an idea reported by the Journal in 2023—the breakneck speed of AI advancement makes that hope ambitious at best. Alphabet reported $50.6 billion in assets not yet in service in 2024, up 44% from a year earlier, while Amazon’s figure rose 62% to $46.4 billion. Microsoft has disclosed fewer details about its own investments.
“It’s a black box in the beginning with the construction-in-progress bucket and then once the project is complete, it’s a black box in the back end,” said Bobby Carnes, an associate professor of clinical accounting at the University of Southern California.
© Arc, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.