The head of Nokia believes the global data centre expansion would accelerate dramatically if manufacturers could overcome current bottlenecks in memory and power infrastructure. Justin Hotard, who previously led Intel's data centre and AI operations before joining Nokia in 2025, contends that the current construction pace does not constitute overbuilding. Meanwhile, consulting firms and economic researchers have begun questioning whether sufficient revenue streams exist to finance the infrastructure being deployed.
Speaking on CNBC's The Tech Download podcast, Hotard stated: "If we could build 2x faster, our customers could build 2x faster, they probably would." He identified memory chips and energy supply as the primary constraints limiting expansion. According to Hotard, this constraint indicates the sector remains in its early phases of growth.
The Nokia executive also addressed concerns about demand sustainability, noting that the sector's growth does not hinge on continuous breakthroughs in artificial intelligence models. He explained: "Even if we didn't have another frontier model released in the next three years, we could probably make tremendous progress just deploying the technology that's there today."
Nokia's business model focuses on connectivity infrastructure rather than semiconductor manufacturing. The company produces equipment that connects server racks within data centres and links facilities to one another. In the second quarter, this division generated EUR 446M in revenue, representing a doubling compared to the prior year, against total group sales of EUR 4.82B. The company's share price has climbed approximately 130% over the past twelve months.
The memory shortage Hotard referenced has already manifested across European markets. In Germany, DDR5 memory pricing surged 414% year-over-year, with memory components now consuming up to half the bill of materials for budget-tier smartphones.
Revenue Gap Concerns
Not all observers share Hotard's optimism regarding demand sustainability. Bain & Company released analysis last week projecting that artificial intelligence infrastructure would require $6T annually by 2031 to support current data centre construction plans. This figure far exceeds the $1.2T to $1.8T that existing products could generate in revenue.
Stijn Van Nieuwerburgh's research presented to the Brookings Papers on Economic Activity estimated total investment requirements at $10.3T between 2025 and 2032, averaging 3.63% of American GDP annually. Van Nieuwerburgh noted that this investment scale would surpass historical infrastructure booms in canals, railways, electrification, highways, and telecommunications relative to economic output.
Van Nieuwerburgh also flagged a structural concern: financing mechanisms are increasingly moving beyond traditional corporate balance sheets into joint ventures, private credit arrangements, and special purpose vehicles.
Europe's Position
Europe has not produced comparable financial assessments of its own infrastructure requirements. Nokia, headquartered in Espoo, operates in a landscape where most of its customer base lies outside the continent.
The European Union launched a bidding process in July for seven AI gigafactories with a combined EUR 30B valuation. However, only approximately EUR 1B in public funding has been formally committed. Meanwhile, the fastest-expanding segment of Nokia's business serves customers engaged in infrastructure development elsewhere.
Source: The Next Web
