As global technology behemoths frantically pour capital into artificial intelligence infrastructure, skeptical voices questioning potential “overbuilding” and an impending “AI bubble” are growing increasingly vociferous across financial markets. However, Nokia CEO Justin Hotard forcefully rebutted this pessimistic narrative during a recent interview with CNBC.
Hotard emphatically argued that the current velocity of expansion cannot be remotely classified as a surplus. Conversely, he asserted that if it were not for crippling shortages in memory chips and electrical power supply, the industry’s pace of data center construction “would be double what it is today.”
Memory and Power: The True Bottlenecks of the AI Gold Rush
Prior to joining Nokia in 2025, Hotard served as the executive vice president and general manager of Intel’s Data Center and AI (DCAI) Group, endowing him with acute, firsthand insight into foundational hardware demands. He underscored that this tidal wave of AI infrastructure deployment remains in its “very early” stages. Even if no revolutionary “frontier models” debut within the next three years, the sheer magnitude of deploying extant AI technologies possesses enough momentum to sustain colossal computational demand.
The authentic culprits obstructing aggressive expansion are severe supply chain constraints and infrastructural bottlenecks. For example, the price of DDR5 memory within Germany has skyrocketed by an astonishing 414% within a single year. Amidst this frenzied AI gold rush, Nokia plays the indispensable role of “selling the shovels.” While they do not manufacture AI processors, they engineer and supply the critical high-speed networking and fiber-optic connectivity equipment linking internal data center racks, as well as bridging discrete data centers globally.
Propelled by this unprecedented construction boom, the specific Nokia division responsible for these technologies successfully doubled its second-quarter sales to €446 million, subsequently catapulting the corporation’s stock price by approximately 130% over the trailing twelve months.
The Looming Revenue Chasm: Warnings from Economists
Nevertheless, this unbridled optimism is not universally shared. Macroeconomists and elite research institutions are issuing stern warnings regarding the ultimate “monetization capacity” of these astronomical AI investments.
Bain & Company recently highlighted a severe financial disparity: to adequately amortize the staggering costs of the AI data centers currently under construction, the technology sector must generate a colossal $6 trillion in annual AI-related revenue by 2031. However, based upon current product commercialization trajectories, they project actual revenues will merely reach between $1.2 trillion and $1.8 trillion, revealing a catastrophic funding chasm.
Furthermore, a comprehensive paper authored by scholar Stijn Van Nieuwerburgh for the Brookings Institution contextualizes this phenomenon historically. He estimates that total AI capital expenditure between 2025 and 2032 will achieve a stratospheric $10.3 trillion, averaging a massive 3.63% of the U.S. GDP annually. He ominously warns that this investment proportion will eclipse the economic footprints of historical infrastructure manias, including the canal, railroad, electrification, and interstate highway booms.
To sustain these astronomical expenditures, immense capital flows are migrating off the balance sheets of tech giants and into joint ventures, private credit markets, and Special Purpose Vehicles (SPVs), subtly inflating systemic risk across the broader financial architecture.
The Collision of Infrastructure Inevitability and Commercial Reality
The profound optimism of Nokia’s CEO and the stark pessimism of macroeconomists are not inherently contradictory; rather, they reflect a critical “temporal disparity” between the hardware supply chain and end-user software commercialization.
From the vantage point of foundational network and compute providers like Nokia or NVIDIA, computational power represents an inelastic, foundational requirement driving productivity for the next decade. Therefore, infrastructure expansion can never proceed too rapidly. However, viewed through the ruthless lens of capital markets, if this $10.3 trillion in Capital Expenditure (CapEx) fails to materialize into tangible, sustainable subscription or service revenues at the consumer or enterprise software level, this highly leveraged hardware arms race—increasingly reliant upon off-balance-sheet financing—will inevitably suffer a brutal and unforgiving valuation correction.
In summation, while AI infrastructure will ultimately democratize technological benefits globally—much like the fiber-optic networks of the dot-com era—a significant cohort of investors, incapable of withstanding the crushing pressure of negative cash flows, will inevitably perish upon the shores before the dawn of true commercial realization.
Support Our Threat Intelligence
If you find our technology report and cybersecurity news helpful, consider supporting our work.
Buy Me a Coffee
PayPal
